POW Software creates innovative & reliable software. Our first solution is ByteSync, a fast and lightweight data synchronization and backup solution. We are also available for freelance C# / .NET development.

ByteSync

This paper is under CC 4.0 license. available on arxiv Authors: (1) Zhe Liu, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (2) Chunyang Chen, Monash University, Melbourne, Australia; (3) Junjie Wang, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China & Corresponding author; (4) Mengzhuo Chen, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (5) Boyu Wu, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (6) Zhilin Tian, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (7) Yuekai Huang, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (8) Jun Hu, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China; (9) Qing Wang, State Key Laboratory of Intelligent Game, Beijing, China Institute of Software Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing, China & Corresponding author. Table of Links Abstract and Introduction Motivational Study and Background Approach Experiment Design Results and Analysis Discussion and Threats to Validity Related Work Conclusion and References 8 CONCLUSION Automated testing is crucial for helping improve app quality. Despite the dozens of mobile app GUI testing techniques, how to automatically generate the diversified unusual text inputs for fully testing mobile apps remains a challenge. This paper proposes InputBlaster which leverages the LLM to produce the unusual inputs together with the mutation rules which serve as the reasoning chains. It formulates the unusual inputs generation problem as a task of producing a set of test generators, each of which can yield a batch of unusual text inputs under the same mutation rule. The evaluation is conducted for both effectiveness and usefulness, with 136% higher bug detection rate than the best baselines, and uncovering 37 new crashes. In the future, we plan to further analyze the root causes and repair strategy of these input-related bugs, and design automated bug repair methods. REFERENCES [1] 2022. Crash bug text. https://www.theguardian.com/technology/iphone-crashbug-text imessage-ios. [2] 2022. Crash bug text in ios. https://tech.hindustantimes.com/tech/news/becareful-a-new-text bomb-is-making-whatsapp-crash-and-will-hang-yourphone-71599532897852.html. [3] 2022. F-droid. https://f-droid.org/. [4] Toufique Ahmed and Premkumar Devanbu. 2022. Few-shot training LLMs for project-specific code-summarization. ASE (2022). [5] Nadia Alshahwan and Mark Harman. 2011. Automated web application testing using search based software engineering. In ASE. IEEE, 3–12. [6] Saswat Anand, Mayur Naik, Mary Jean Harrold, and Hongseok Yang. 2012. Automated concolic testing of smartphone apps. In Proceedings of the ACM SIGSOFT 20th International Symposium on the Foundations of Software Engineering. 1–11. [7] Yauhen Leanidavich Arnatovich, Minh Ngoc Ngo, Tan Hee Beng Kuan, and Charlie Soh. 2016. Achieving high code coverage in android ui testing via automated widget exercising. In 2016 23rd Asia-Pacific Software Engineering Conference (APSEC). IEEE, 193–200. [8] Yauhen Leanidavich Arnatovich, Lipo Wang, Ngoc Minh Ngo, and Charlie Soh. 2018. Mobolic: An automated approach to exercising mobile application GUIs using symbiosis of online testing technique and customated input generation. Software: Practice and Experience 48, 5 (2018), 1107 1142. [9] Tanzirul Azim and Iulian Neamtiu. 2013. Targeted and depth-first exploration for systematic testing of android apps. In Proceedings of the 2013 ACM SIGPLAN international conference on Object oriented programming systems languages & applications. 641–660. [10] Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901. [11] Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901. [12] Tianqin Cai, Zhao Zhang, and Ping Yang. 2020. Fastbot: A Multi-Agent ModelBased Test Generation System Beijing Bytedance Network Technology Co., Ltd.. In Proceedings of the IEEE/ACM 1st International Conference on Automation of Software Test. 93–96. [13] Chen Chen, Baojiang Cui, Jinxin Ma, Runpu Wu, Jianchao Guo, and Wenqian Liu. 2018. A systematic review of fuzzing techniques. Computers & Security 75 (2018), 118–137. [14] Taolue Chen, Alejandro Flores-Lamas, Matthew Hague, Zhilei Han, Denghang Hu, Shuanglong Kan, Anthony W Lin, Philipp Rümmer, and Zhilin Wu. 2022. Solving string constraints with Regex dependent functions through transducers with priorities and variables. Proceedings of the ACM on Programming Languages 6, POPL (2022), 1–31. [15] Taolue Chen, Matthew Hague, Jinlong He, Denghang Hu, Anthony Widjaja Lin, Philipp Rümmer, and Zhilin Wu. 2020. A decision procedure for path feasibility of string manipulating programs with integer data type. In Automated Technology for Verification and Analysis: 18th International Symposium, ATVA 2020, Hanoi, Vietnam, October 19–23, 2020, Proceedings. Springer, 325–342. [16] Ting Chen, Simon Kornblith, Kevin Swersky, Mohammad Norouzi, and Geoffrey E Hinton. 2020. Big self-supervised models are strong semi-supervised learners. Advances in neural information processing systems 33 (2020), 22243–22255. [17] Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, et al. 2022. Palm: Scaling language modeling with pathways. arXiv preprint arXiv:2204.02311 (2022). [18] Joel D Day, Thorsten Ehlers, Mitja Kulczynski, Florin Manea, Dirk Nowotka, and Danny Bøgsted Poulsen. 2019. On solving word equations using SAT. In Reachability Problems: 13th International Conference, RP 2019, Brussels, Belgium, September 11–13, 2019, Proceedings 13. Springer, 93–106. [19] Biplab Deka, Zifeng Huang, Chad Franzen, Joshua Hibschman, Daniel Afergan,  Yang Li, Jeffrey Nichols, and Ranjitha Kumar. 2017. Rico: A Mobile App Dataset for Building Data-Driven Design Applications. In UIST. [20] Yinlin Deng, Chunqiu Steven Xia, Chenyuan Yang, Shizhuo Dylan Zhang, Shujing Yang, and Lingming Zhang. 2023. Large Language Models are Edge-Case Fuzzers: Testing Deep Learning Libraries via FuzzGPT. ISSTA (2023). [21] Yinlin Deng, Chenyuan Yang, Anjiang Wei, and Lingming Zhang. 2022. Fuzzing deep-learning libraries via automated relational API inference. In Proceedings of the 30th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering. 44–56. [22] Android Developers. 2012. Ui/application exerciser monkey. [23] Zhen Dong, Marcel Böhme, Lucia Cojocaru, and Abhik Roychoudhury. 2020. Time-travel testing of android apps. In 2020 IEEE/ACM 42nd International Conference on Software Engineering (ICSE). IEEE, 481–492. [24] Robert Feldt and Simon Poulding. 2013. Finding test data with specific properties via metaheuristic search. In 2013 IEEE 24th International Symposium on Software Reliability Engineering (ISSRE). IEEE, 350–359. [25] Shivam Garg, Dimitris Tsipras, Percy S Liang, and Gregory Valiant. 2022. What can transformers learn in-context? a case study of simple function classes. Advances in Neural Information Processing Systems 35 (2022), 30583–30598. [26] Tianxiao Gu, Chengnian Sun, Xiaoxing Ma, Chun Cao, Chang Xu, Yuan Yao, Qirun Zhang, Jian Lu, and Zhendong Su. 2019. Practical GUI testing of Android applications via model abstraction and refinement. In 2019 IEEE/ACM 41st International Conference on Software Engineering (ICSE). IEEE, 269–280. [27] Yuyu He, Lei Zhang, Zhemin Yang, Yinzhi Cao, Keke Lian, Shuai Li, Wei Yang, Zhibo Zhang, Min Yang, Yuan Zhang, et al. 2020. TextExerciser: feedback-driven text input exercising for android applications. In 2020 IEEE Symposium on Security and Privacy (SP). IEEE, 1071–1087. [28] Luk Holk, Petr Jank, Anthony W Lin, and Rmmer. 2017. String constraints with concatenation and transducers solved efficiently. Proceedings of the ACM on Programming Languages 2, POPL (2017), 1–32. [29] Yang Hu, Umair Z Ahmed, Sergey Mechtaev, Ben Leong, and Abhik Roychoudhury. 2019. Re factoring based program repair applied to programming assignments. In 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE). IEEE, 388–398. [30] Text input. 2022. Introduction about text input on Android Developer website. https://developer.android.google.cn/reference/android/widget/EditText?hl=en. [31] Nan Jiang, Kevin Liu, Thibaud Lutellier, and Lin Tan. 2023. Impact of Code Language Models on Automated Program Repair. ICSE (2023). [32] Sungmin Kang, Juyeon Yoon, and Shin Yoo. 2023. Large Language Models are Few-shot Testers: Exploring LLM-based General Bug Reproduction. ICSE (2023). [33] Adam Kiezun, Vijay Ganesh, Shay Artzi, Philip J Guo, Pieter Hooimeijer, and Michael D Ernst. 2013. HAMPI: A solver for word equations over strings, regular expressions, and context-free grammars. ACM Transactions on Software Engineering and Methodology (TOSEM) 21, 4 (2013), 1–28. [34] Sebastian Krings, Joshua Schmidt, Patrick Skowronek, Jannik Dunkelau, and Dierk Ehmke. 2020. Towards constraint logic programming over strings for test data generation. In Declarative Programming and Knowledge Management: Conference on Declarative Programming, DECLARE 2019, Unifying INAP, WLP, and WFLP, Cottbus, Germany, September 9–12, 2019, Revised Selected Papers 22. Springer, 139–159. [35] Misha Laskin, Kimin Lee, Adam Stooke, Lerrel Pinto, Pieter Abbeel, and Aravind Srinivas. 2020. Reinforcement learning with augmented data. Advances in neural information processing systems 33 (2020), 19884–19895. [36] Caroline Lemieux, Jeevana Priya Inala, Shuvendu K Lahiri, and Siddhartha Sen. 2023. CODAMOSA: Escaping Coverage Plateaus in Test Generation with Pretrained Large Language Models. In ICSE. [37] Guodong Li and Indradeep Ghosh. 2013. PASS: String solving with parameterized array and interval automaton. In Hardware and Software: Verification and Testing: 9th International Haifa Verification Conference, HVC 2013, Haifa, Israel, November 5-7, 2013, Proceedings 9. Springer, 15–31. [38] Yuanchun Li, Ziyue Yang, Yao Guo, and Xiangqun Chen. 2017. DroidBot: A Lightweight UI-Guided Test Input Generator for Android (ICSE-C ’17). [39] Yuanchun Li, Ziyue Yang, Yao Guo, and Xiangqun Chen. 2017. Droidbot: a lightweight ui-guided test input generator for android. In 2017 IEEE/ACM 39th International Conference on Software Engineering Companion (ICSE-C). IEEE, 23– 26. [40] Yuanchun Li, Ziyue Yang, Yao Guo, and Xiangqun Chen. 2019. Humanoid: a deep learning-based approach to automated black-box Android app testing. In ASE. IEEE, 1070–1073. [41] Hongliang Liang, Xiaoxiao Pei, Xiaodong Jia, Wuwei Shen, and Jian Zhang. 2018. Fuzzing: State of the art. IEEE Transactions on Reliability 67, 3 (2018), 1199–1218. [42] Tianyi Liang, Andrew Reynolds, Cesare Tinelli, Clark Barrett, and Morgan Deters. 2014. A DPLL (T) theory solver for a theory of strings and regular expressions. In Computer Aided Verification: 26th International Conference, CAV 2014, Held as Part of the Vienna Summer of Logic, VSL 2014, Vienna, Austria, July 18-22, 2014. Proceedings 26. Springer, 646–662. [43] Peng Liu, Xiangyu Zhang, Marco Pistoia, Yunhui Zheng, Manoel Marques, and Lingfei Zeng. 2017. Automatic text input generation for mobile testing. In 2017 IEEE/ACM 39th International Conference on Software Engineering (ICSE). IEEE, 643–653. [44] Zhe Liu, Chunyang Chen, Junjie Wang, Xing Che, Yuekai Huang, Jun Hu, and Qing Wang. 2022. Fill in the Blank: Context-aware Automated Text Input Generation for Mobile GUI Testing. arXiv preprint arXiv:2212.04732 (2022). [45] Zhe Liu, Chunyang Chen, Junjie Wang, Yuekai Huang, Jun Hu, and Qing Wang. 2020. Owl Eyes: Spotting UI Display Issues via Visual Understanding. In ASE. IEEE. https://doi.org/10.1145/3324884.3416547 [46] Zhe Liu, Chunyang Chen, Junjie Wang, Yuekai Huang, Jun Hu, and Qing Wang. 2022. Nighthawk: Fully Automated Localizing UI Display Issues via Visual Understanding. IEEE Transactions on Software Engineering (2022), 1–16. https: //doi.org/10.1109/TSE.2022.3150876 [47] Li Lucy and David Bamman. 2021. Gender and representation bias in GPT-3 generated stories. In Proceedings of the Third Workshop on Narrative Understanding. 48–55. [48] Aravind Machiry, Rohan Tahiliani, and Mayur Naik. 2013. Dynodroid: An input generation system for android apps. In Proceedings of the 2013 9th Joint Meeting on Foundations of Software Engineering. 224–234. [49] Ke Mao, Mark Harman, and Yue Jia. 2016. Sapienz: Multi-objective automated testing for Android applications. In Proceedings of the 25th International Symposium on Software Testing and Analysis. 94–105. [50] Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Efficient Estimation of Word Representations in Vector Space. Computer Science (2013). [51] Sewon Min, Xinxi Lyu, Ari Holtzman, Mikel Artetxe, Mike Lewis, Hannaneh Hajishirzi, and Luke Zettlemoyer. 2022. Rethinking the Role of Demonstrations: What Makes In-Context Learning Work? arXiv preprint arXiv:2202.12837 (2022). [52] Noor Nashid, Mifta Sintaha, and Ali Mesbah. 2023. Retrieval-Based Prompt Selection for Code Related Few-Shot Learning. In Proceedings of the 45th International Conference on Software Engineering (ICSE’23). [53] Minxue Pan, An Huang, Guoxin Wang, Tian Zhang, and Xuandong Li. 2020. Reinforcement learning based curiosity-driven testing of Android applications. In Proceedings of the 29th ACM SIGSOFT International Symposium on Software Testing and Analysis. 153–164. [54] Gabriel Poesia, Oleksandr Polozov, Vu Le, Ashish Tiwari, Gustavo Soares, Christopher Meek, and Sumit Gulwani. 2022. Synchromesh: Reliable code generation from pre-trained language models. ICLR (2022). [55] Simon Poulding and Robert Feldt. 2017. Generating controllably invalid and atypical inputs for robustness testing. In 2017 IEEE International Conference on Software Testing, Verification and Validation Workshops (ICSTW). IEEE, 81–84. [56] Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer. J. Mach. Learn. Res. 21 (2020), 140:1–140:67. [57] Vaibhav Rastogi, Yan Chen, and William Enck. 2013. Appsplayground: automatic security analysis of smartphone applications. In Proceedings of the third ACM conference on Data and application security and privacy. 209–220. [58] J Schulman, B Zoph, C Kim, J Hilton, J Menick, J Weng, JFC Uribe, L Fedus, L Metz, M Pokorny, et al. 2022. ChatGPT: Optimizing language models for dialogue. [59] Carolyn B. Seaman. 1999. Qualitative methods in empirical studies of software engineering. IEEE Transactions on software engineering 25, 4 (1999), 557–572. [60] Mike Sharples. 2022. Automated Essay Writing: An AIED Opinion. International Journal of Artificial Intelligence in Education (2022), 1–8. [61] Ting Su, Guozhu Meng, Yuting Chen, Ke Wu, Weiming Yang, Yao Yao, Geguang Pu, Yang Liu, and Zhendong Su. 2017. Guided, stochastic model-based GUI testing of Android apps. In Proceedings of the 2017 11th Joint Meeting on Foundations of Software Engineering. 245–256. [62] Nezih Sunman, Yiğit Soydan, and Hasan Sözer. 2022. Automated web application testing driven by pre-recorded test cases. Journal of Systems and Software (2022), 111441. [63] Minh-Thai Trinh, Duc-Hiep Chu, and Joxan Jaffar. 2014. S3: A symbolic string solver for vulnerability detection in web applications. In Proceedings of the 2014 ACM SIGSAC Conference on Computer and Communications Security. 1232–1243. [64] Minh-Thai Trinh, Duc-Hiep Chu, and Joxan Jaffar. 2017. Model counting for recursively-defined strings. In Computer Aided Verification: 29th International Conference, CAV 2017, Heidelberg, Germany, July 24-28, 2017, Proceedings, Part II 30. Springer, 399–418. [65] UIAutomator. 2021. Python wrapper of Android uiautomator test tool. https://github. com/xiaocong/uiautomator. [66] Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems (2017). [67] Jue Wang, Yanyan Jiang, Chang Xu, Chun Cao, Xiaoxing Ma, and Jian Lu. 2020. Combodroid: generating high-quality test inputs for android apps via use case combinations. In ICSE. 469–480. [68] Zhengyuan Yang, Zhe Gan, Jianfeng Wang, Xiaowei Hu, Yumao Lu, Zicheng Liu, and Lijuan Wang. 2022. An empirical study of gpt-3 for few-shot knowledgebased vqa. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 36. 3081–3089. [69] Zhengran Zeng, Hanzhuo Tan, Haotian Zhang, Jing Li, Yuqun Zhang, and Lingming Zhang. 2022. An extensive study on pre-trained models for program understanding and generation. In Proceedings of the 31st ACM SIGSOFT International Symposium on Software Testing and Analysis. 39–51. [70] Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui  Chen, Christopher Dewan, Mona Diab, Xian Li, Xi Victoria Lin, et al. 2022. Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068 (2022).

Leveraging LLMs for Generation of Unusual Text Inputs in Mobile App Tests: Conclusion and References

Hey hackers! Did you know that HackerNoon runs a to gauge reader sentiment on topics of interest? That’s right, every Monday readers are greeted with a new poll curated by a HackerNoon editor on some of the most trending news on the web (plus some general, wacky ones too just for fun!). weekly poll Now all that’s great, but what if we told you that there’s a way to go a step further with these polls and truly analyze readers’ sentiment? Let’s look at the : current poll running on HackerNoon Poll #1 Lesson: How to vote? This week's poll focuses on the debate over the potential ban of ByteDance’s TikTok in the US over national security concerns. Is it an overreach by the US government or a step in the right direction? To vote, pick the option that best aligns with your thoughts and compare it with the broader reader population 😎 Poll #2 Lesson: How to check past polls? Click on the “ ” to access a list of all the polls HackerNoon has run in the past. Poll of the Week Now, select a poll that has already closed and analyze the results 🥳 Poll #3 Lesson: How do I access polls' templates? Let’s look at the “ ” poll: if you click on the poll, it should show you a prompt that you can use to start analyzing the results right away! Take a look: What is your web browser of choice Got it?! AWESOME 👏 Here are some of our favorite polls that are definitely worthy of your analysis: How Much Faith Do You Have in Your Government to Effectively Regulate AI? Will AI stocks suffer the same fate as the dot-com bubble? Where do YOU think bitcoin's price is heading by the end of the year? Do You Think the "Metaverse" Will Ever Truly Take Off? Other News Around The Web Nvidia GTC Ah, Nvidia. What can be said about the company that hasn’t been said already after it became the arms dealer of choice for the generative artificial intelligence boom. Well, apparently a lot! The company its annual GPU Technology Conference, or GTC, heralding the launch of the Blackwell class of AI chips that are supposed to be better than their Hopper predecessors. But what could these new processors mean for the development of more powerful AI and what does Nvidia’s change in business direction mean for consumers at large? recently held Let us know ! here TikTok’s Future in the US Should TikTok face a ban in the US? That’s the question we’re readers this week as we highlight this important issue. The US House recently passed a bill requiring ByteDance, the parent company of TikTok, to divest from the app within 180 days or face prohibition (a ban) in the U.S. However, Senate supporters argue that this is not a ban but rather a transition of ownership to ensure national security. Is this overreaction, and perhaps, an overreach by the US government? asking and share your thoughts . Vote in the Poll here Sam Bankman-(never getting)Freed Looks like Sam Bankman-Fried’s days as a billionaire and crypto entrepreneur are over (not that there was ever a chance he would be able to dazzle in crypto ever again anyway). The once wildly successful owner of crypto exchange FTX learned this past week what financial mismanagement gets you; a quarter of a century in prison. SBF’s sentencing, which he’s going to appeal, apparently also saw a wild increase in the prices of memecoins (we’re talking thousands of %age up). In light of what’s happened, what do you of the sentencing and how do you in 2024? think feel about cryptocurrencies Optout Writing Contest HackerNoon’s partnership with Aut Labs is still going strong and we’ve seen some great entries these past few weeks! If you’d like to take a stab at thousands of dollars in USDT, simply head on down to and start writing on the Contest and tell us how YOU are opting out of a system that does not represent you. winning our template here #Optout Happy writing!

How to Use HackerNoon Polls to Your Advantage 🤔

আপনার সুবিধার জন্য হ্যাকারনুন পোল কীভাবে ব্যবহার করবেন 🤔

So nutzen Sie HackerNoon-Umfragen zu Ihrem Vorteil 🤔

Cómo utilizar las encuestas de HackerNoon a su favor 🤔

Comment utiliser les sondages HackerNoon à votre avantage 🤔

HackerNoon पोल का अपने लाभ के लिए उपयोग कैसे करें 🤔

HackerNoon のアンケートを有効活用する方法 🤔

HackerNoon 설문조사를 유리하게 활용하는 방법 🤔

Como usar as pesquisas HackerNoon a seu favor 🤔

Как использовать опросы HackerNoon в своих интересах 🤔

HackerNoon Anketlerini Avantajınıza Nasıl Kullanabilirsiniz 🤔

Cách sử dụng HackerNoon Polls để mang lại lợi ích cho bạn 🤔

如何利用 HackerNoon 民意调查为自己谋利 🤔

Sam Bankman-Fried has been sentenced to a quarter of a century in prison after he and his lawyers failed to convince a U.S. district judge that the former crypto mogul was an autistic altruist who never intended to harm anybody. The judgment itself is pretty damning, and while SBF's lawyers plan to appeal, there's a lot to unpack here. First, regardless of how the appeal process plays out, Sam Bankman is probably going to prison for a long, long time. And despite attempts at maintaining his innocence, the prosecution's evidence points to the contrary. Second, Sam Bankman-Fried's fall from grace is not so much about what happened but it happened. For those that don't remember, everything was going A-OK at Sam Bankman-Fried's crypto exchange, FTX, until he decided to pick a fight with Binance founder and former Chief Executive Changpeng "CZ" Zhao which led the latter to (allegedly) cause a bank run that resulted in the demise of the third-largest crypto exchange in the US. how Third, there is something inherently wrong with the crypto industry, as it always seems to attract some of the most dubious personalities in the world. Speaking of which, Binance's Zhao may have had the last laugh following the FTX debacle but his victory has been pretty short-lived since he has had to step down as the CEO of the world's largest crypto exchange. This decision came as part of a guilty plea with the US Department of Justice over a number of charges, including money laundering. The only exchange that has been able to weather most, if not all, of the turbulence so far has been Coinbase and its Chief Executive Brian Armstrong. Look, I don't know what it is with the Lex Luthor lookalike, but Armstrong clearly knows how to navigate through these troubling times to come out on top. Anyway, the internet did what it always does when things like these happen: pump and dump memecoins, leaving some with big wads of cash and most with a worthless token. Now, if you're wondering whether the sentence is fair, first of all, nothing in life is fair, but second: a lot of the agree that things have looked differently for SBF if he had actually managed to show some actual remorse over what happened and owned up to his mistakes. media reports might But since SBF kept skirting around the issue, and the judge in the case noticed, he's been sentenced to prison for two-and-a-half decades. I'll keep you posted on the appeal proceedings. Sorry to break your immersion, hackers. We have an important update to share. The Tech Company News Brief aka the newsletter you’re currently reading is going to go through an overhaul in the near future with a new theme and a new name. Keep your eyes peeled! We’re very excited to show you the new format (soon!). Amazon to Say Buh-Bye to Cashierless System 💸 If you were hoping to hop in-n-out of an store in the US, I've got bad news for you. Amazon The world's foremost authority on everything eCommerce has decided that it wants nothing to do with the technology (which it lovingly called Just Walk Out), at least in the US. According to , Amazon will be stamping out the system from its Fresh grocery stores and replacing it with the Dash Cart instead which allows buyers to scan items as they go. media reports Amazon ranked #9 on HackerNoon's Tech Company Rankings this week. 👋 You’re reading HackerNoon's , a weekly collection of tech goodness that combines HackerNoon's proprietary data with internet trends to determine which companies are rising and falling in the public consciousness. Subscribe to receive the newsletter in your inbox every Tuesday. Tech Company News Brief here complete Canada Comes After Social Media Big Boys The good folks in Canada have had it with social media companies' shenanigans, so much so that they're taking them to court. The Toronto District School Board has filed a lawsuit against Meta Platforms, TikTok owner ByteDance, and Snapchat for allegedly creating addictive products that are harmful to a younger audience (think adolescents and teens) and has sought 4.5 billion Canadian dollars in damages (roughly $3.3 billion US). This wouldn't be the first time social media companies are blamed for a decline in youth mental health, with research, both independent and commissioned by these companies themselves, showing a clear link between the two. Whether the companies actually end up paying for the damage they've caused (or might be causing) is yet to be seen. Meta's and ranked #2 and #4 this week, respectively. Facebook Instagram In Other News.. 📰 Bitcoin ETF Trading Volumes Tripled in March as the Largest Cryptocurrency Hit Record Highs — via CoinDesk Opera allows users to download and use LLMs locally — via TechCrunch US government review faults Microsoft for ‘cascade’ of errors that allowed Chinese hackers to breach senior US officials’ emails — via CNN Microsoft, Quantinuum claim breakthrough in quantum computing — via Reuters AI security is a new battle between employers and workers, survey shows — via Axios Zoom and Salesforce stand to benefit from Microsoft’s unbundling of Teams and Office — via CNBC And that's a wrap! Don't forget to share this newsletter with your family and friends! See y'all next week. PEACE! ☮️ — Sheharyar Khan, Editor, Business Tech @ HackerNoon *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's page. Tech Company Rankings

Sam Bankman-(Never Getting)Fried

স্যাম ব্যাংকম্যান- (কখনও পাওয়া যায় না) ভাজা

Sam Bankman - (Never Getting) Fried

Sam Bankman-(Nunca conseguir) frito

Sam Bankman-(Ne jamais obtenir)Fried

सैम बैंकमैन-(नेवर गेटिंग)फ्राइड

サム・バンクマン - (決して) 揚げられない

샘 뱅크맨(Sam Bankman) - (Never Getting)Fried

Sam Bankman- (Nunca Ficando) Frito

Сэм Бэнкман - (Никогда не получается) Жареный

Sam Bankman-(Asla Kızartılmaz)

Over the last couple of days, I have seen many of my marketer colleagues and AI researchers debating over the potential ban and defending the platform blindly. TikTok As someone who not only runs ad campaigns on social media and has attended human-centered computer design courses for years but also spent some time in libertarian organizations defending the free market and individualism over state control, I have a few insights. Let me be unequivocally clear: my advocacy for AI, free from government control, is mirrored in my stance on social media platforms. However, the crux of the current debate focuses on China's proven exploitation of TikTok to surveil Western people, leaders as well as journalists and shape the ideological trajectory of our youth. open-source Without delving too deeply into politics, let me outline with data and research papers why I strongly support the bipartisan bill to ban TikTok, and why you should delete the app now! Before TikTok, social media was social Instagram, Twitter, and Snapchat; these apps focused on interactions between friends, groups, or communities until TikTok brought the dopamine hit and sucked users' attention spans. The "Likes and Fragments: Examining Perceptions of Time Spent on TikTok" research is a proven example of how TikTok users . often misjudge their usage time Findings highlight that users who engaged more intensively with the platform by liking more videos and opening the app more frequently had a higher discrepancy between their perceived and actual usage times. This suggests that the engaging and immersive nature of TikTok, combined with its algorithmically curated "For You" feed, creates an environment where users are likely to misjudge the amount of time they spend on the app. This is important because it leads us to TikTok's algorithm, which heavily relies on casino-like tactics. You have to be over 18 or 21 to gamble, but this app gets the next generations addicted from a young age. So, what exactly are these tactics? The Devious Tactics of TikTok's Algorithm Significant research from the Brown University School of Public Health highlights the addictive nature of TikTok. The design of TikTok uses classical conditioning and reward-based learning processes to encourage continuous use. Features such as the "For You" stream customize content feeds based on user interactions. Moreover, TikTok employs a variable , with the "swipe down" feature to refresh the feed mirroring the action of pulling a slot machine lever. This intermittent reinforcement keeps users engaged, always anticipating the next "rewarding" video. reward pattern similar to slot machines I have explored the algorithm below more. This strategic manipulation exploits classical conditioning and reward-based learning processes, making the platform highly addictive and challenging to resist. Last year, Utah Attorney General Sean Reyes filed a lawsuit against TikTok, showing how the "For You" recommendation algorithm hooks young users. You can read the here for further exploration. lawsuit showing TikTok's casino tactics “TikTok induces that effect by relying on the same strategy casinos use to hook gamblers on slot machines – keeping them at the machine, spending a quarter at a time, hoping that the next pull will be ‘the one,’” These random dopamine releases and other manipulative tactics hook the not-yet-fully-developed brains of our youth to this state-owned media channel. Now, let's discuss why we should not leave the control of our next generation's brain development to communist China. The Cultural Dichotomy A comparison between Douyin (Chinese version of TikTok) and ours after China's National Intelligence took control of ByteDance shows a massive difference. In 2021, China's National Intelligence . They appointed a government official with a golden vote (this person has the final say in the company), and after this change, Douyin's algorithms entirely changed. took over the control of ByteDance While your children are bombarded with 'trash content', Chinese generations are exposed to . While your one and only child spends hours doom scrolling, their children have mandatory time limits. educational content Marketing luminary Scott Galloway gives a : strong reason to avoid TikTok "If I were part of the CCP and recognized a vested interest in undermining America's strategic position globally, I would subtly and cunningly tip the balance of content to cast America in an unfavorable light. I believe this is exactly what's happening at this moment—it would be foolish for them not to engage in such tactics." The Addiction and Its Impacts An average TikTok user spends about on the app and it is projected to exceed 100 minutes this year. The hyper-personalization algorithm leads to addiction-like behaviors. 60 minutes a day It has been proven repeatedly that TikTok's hyper-personalization algorithm leads to problematic use patterns similar to . addiction-like behaviors TikTok encourages isolation and loneliness among young people, and year-over-year depression statistics show that following the widespread adoption of this soft power weapon. depression among youth has significantly increased "Individuals, particularly young people with social anxiety or depression, might be inclined to spend excessive time online, diminishing real-life interactions," explains Anne Marie Albano, PhD, director of the Columbia University Clinic for Anxiety and Related Disorders. "Avoiding active engagement with the world — through social exchanges, confronting challenges, participating in discussions, or navigating peer conflicts — can intensify feelings of isolation, despair, and emotional distress." The findings reveal that using TikTok elicits negative emotions in users, affecting their intention to continue using these platforms. It leads to feelings of wasted time, addiction, and even impacts mental health when usage becomes excessive or disrupts daily life. research proves that Tiktok's leads to consuming more and ultimately, addiction. "Understanding users’ negative emotions and continuous usage intention in short video platforms" negative emotion flow algorithm Weaponizing TikTok TikTok serves as a weapon for the state-sponsored agenda. The research conducted by the Network Contagion Research Institute in collaboration with Rutgers University Miller Center investigates to the Chinese Government. discrepancies in hashtag ratios on sensitive topics For non-sensitive topics, such as general politics and pop culture, the hashtag ratios between Instagram and TikTok approximately followed the user ratios (~2:1). This pattern of anomalies was consistent across national/regional and international issues sensitive to China, suggesting a systematic approach by TikTok to influence content visibility based on its alignment with Chinese Government interests. NCRI's research extends to geopolitically sensitive topics, demonstrating underrepresentation of hashtags , and overrepresentation of those in alignment, particularly regarding international conflicts where China has strategic interests. favoring narratives opposed to Chinese Government interests The findings raise concerns about TikTok's impartiality and the platform's potential use in furthering the national/regional or international objectives of the Chinese Government. The discrepancies in hashtag representation underscore the need for a more comprehensive analysis to understand TikTok's influence on public narratives and its alignment with the Chinese Government's geostrategic objectives. The NCRI's work highlights the importance of transparency and scrutiny in digital platforms' operations, especially when they have a significant impact on public discourse and possess the potential for governmental influence. Indeed, it is a tool utilized for advancing state-sponsored agendas, with the recent attributed to bin Laden being just one of the more notorious examples. "Letter to America" trend There are two primary uses of this instrument: and . extensive political manipulation concerns regarding data security It has been established that to its younger audience. A Wall Street Journal investigation reveals that TikTok's algorithm has displayed footage of explosions, rockets, and distressed families from war zones to a 13-year-old user who has just signed up, highlighting the platform's exposure of young users to such graphic content. TikTok amplifies content related to self-harm, mental health issues, and extremism Furthermore, research by Newsguard indicates that spending as little as 40 minutes on TikTok leads to exposure to , such as Russia's invasion of Ukraine. false content regarding political events The study " examines how user behaviors and settings affect the recommendations made by TikTok's algorithm. An Empirical Investigation of Personalization Factors on TikTok" The researchers used controlled "sock-puppet" accounts to focus on the impact of personalization factors, including language preference, geographical location, user interactions such as likes and follows, and video viewing durations, on the content that TikTok suggests to its users. The with limited control over the content they are shown leads to echo chambers and filter bubbles, where users are increasingly exposed to content that reinforces their existing beliefs and interests, limiting exposure to diverse viewpoints. "black box" nature of TikTok's recommendation algorithm These researches are showing that this app doesn't only make you addicted while diminishing your ability to pay attention, but it also deliberately changes the future of our very own civilization. The other concern pertains to data security. With over on iPhones and Android devices in the USA, there are nearly 112 million distinct data sources, encompassing not only content but also real-time voice streams. 112 million TikTok installations Data Security The by Zhuoran Liu et al. at ByteDance Inc. showcases ByteDance's position in . "Monolith: Real-Time Recommendation System With Collisionless Embedding Table" leveraging complex algorithms This is a system engineered for real-time recommendation. Monolith addresses critical challenges inherent in recommendation systems, such as handling sparse, categorical data and adapting to non-stationary user behavior, through solutions like a collisionless embedding table and dynamic feature eviction mechanisms. Let's clarify this; Traditional recommendation systems often rely on batch training, where data is collected over a period, processed, and then used to update the model's recommendations. This can lead to delays in the model reflecting the latest user interactions. Monolith's system, by contrast, uses online training. This means it learns from user interactions in real-time, allowing for a much more responsive and personalized user experience. The user's interaction with content is immediately fed back into the system, enabling it to adjust its recommendations on the fly. In the context of machine learning, an embedding table transforms sparse, categorical data into dense vectors of fixed size, which the model can more easily work with. A "collision" in this context would mean different input data points being mapped to the same vector representation, which could lead to inaccurate model predictions. A collisionless embedding table suggests a method where these collisions are minimized or eliminated, improving the model's accuracy and efficiency in handling vast amounts of categorical data. The system's architecture utilizes online training, enabling immediate adaptation based on user interactions, thus significantly refining content personalization. However, this state-of-the-art AI personalization raises important discussions about data privacy, as the continuous, involves , often without or clear data usage policies. real-time processing of user data for personalized content delivery real-time extensive data collection and analysis transparent consent mechanisms . TikTok is under scrutiny for the extent of data it gathers to customize content and refine its algorithm. There is concern that this data could potentially be accessed by the authoritarian government and used against us. One might counter-argue that ByteDance recently proposed transferring their . However, accepting this at face value is not only naive but also highlights that ensuring data security is an almost impossible task. data centers to the US by using Oracle servers The Urgent Need for Action China has been exposing its plans to since 2017. rule the world with artificial general intelligence Deleting TikTok from our phones, creating a public narrative against its use, and sharing these findings and research to enlighten our friends are crucial steps. I believe I am doing my part by writing this article to showcase why we need immediate action against this state-controlled app. "Control over the use of one's mind is integral to human autonomy, and denial of such control undermines freedom of thought and freedom of will." To gain deeper insights into the digital competition and its critical implications for the future, I highly recommend tuning in to the session, if you haven't watched it yet. All-In Summit: Bill Gurley 2,851 Miles Here, Gurley delves into discussions about regulations, the future of AI, and the upcoming intense historical crash between China and the USA. I urge you to share this content with your friends who use TikTok and encourage them to take immediate action by deleting this spyware app. Your participation in this conversation is crucial. Feel free to share your thoughts and comments below, as we continue to explore and address this important topic together. Also published here.

TikTok Should Have Been Banned Years Ago, and Here's Why

TikTok বছর আগে নিষিদ্ধ করা উচিত ছিল, এবং এখানে কেন

TikTok hätte schon vor Jahren verboten werden sollen, und hier ist der Grund

TikTok debería haber sido prohibido hace años, y este es el motivo

TikTok aurait dû être interdit il y a des années, et voici pourquoi

टिकटॉक पर वर्षों पहले ही प्रतिबंध लग जाना चाहिए था, और यहां बताया गया है क्यों

TikTokは何年も前に禁止されるべきだった、その理由は次のとおりです

TikTok은 몇 년 전에 금지되었어야 했는데, 그 이유는 다음과 같습니다.

O TikTok deveria ter sido banido há anos e aqui está o porquê

TikTok следовало запретить много лет назад, и вот почему

TikTok Yıllar Önce Yasaklanmalıydı ve İşte Nedeni

Lẽ ra TikTok đã bị cấm từ nhiều năm trước và đây là lý do

TikTok 几年前就应该被禁止，原因如下

Hey everyone! and just like you, I’ve been fascinated with the recent progress of artificial intelligence. Realizing that I needed to stay abreast with all the developments happening, I decided to embark on a personal journey of learning, thus was born! With this series, I will be learning about LLMs and share ideas, experiments, opinions, trends & learnings through my blog posts. You can follow along the journey on HackerNoon or my personal website . In today’s article, we’ll be looking to build a Semantic Kernel with the help of GPT-4. , I’m Nataraj 100 days of AI here here My goal with this post is to look at how startups at different stages leveraged LLMs and understand how they integrated AI into their products. The goal of this exercise is to come up with best practices that startups can learn from these examples. Example 1: Notion AI In a classic startup fashion the first version of Notion AI was coded in a hotel room during a company retreat in couple of days. Timeline – Private alpha in November 2022. We spent 10 weeks gathering vital user feedback before February 2023, when Notion AI became generally available to tens of millions of users Features launched: Completion of blog ideas to create first draft. Generate ideas to accomplish a task. Check grammar check, translation to different languages & Summarize a document. Help me write, to unblock writers and get them started. Example 2: Stripe Stripe took a drastic step and asked 100 of its employees to stop what they are doing and come up with ideas on how to use GPT4 to make stripe better. The result was a whole new set of features. Launched a way to summarize their customer business information, which was better than what people would write. Launched a chat bot to answer questions about its technical documentation, improves developer & partner productivity. Used GPT4 to find malicious accounts in their discord community. Example 3: Shopify Shopify was quick to adopt gen AI and launched a suite of AI features under the product name Shopify Magic. Shopify Magic is a collection of gen AI features which include: Text Generation – Speed up writing product descriptions, blog posts, email subject lines, email replies & headlines in Shopify stores. Basically anywhere a Shopify store admin has to write text they have added text generation capability. Media Generation – Product images play a big role in ecommerce. Editing image with out using any other apps is now added. Summarize App review – Instead of looking at 100’s of app reviews to find the right app to install you can look at summarized app reviews to find the right app. Example 4: CapCut If you have not heard of CapCut before, it is a video editing tool created by TikTok’s parent company Bytedance to help creators on TikTok. I use it to edit my podcast on a regular basis and I think its the most easiest & advanced editing tool out there for content creators. CapCut is also constantly updating its software and introducing new AI features. Here are some of them. Generate subtitles with one click. Generate videos with a text or image as a prompt and also give an ability to edit them once they are generated. The most latest feature they launched is CapCut will automatically generate shorts for you which you can further edit from a long form video. Although its not perfect it does save hours if you are an editor. Example 5: Yabble Yabble is a tool that delivers insights for business using which they can make important decisions. With lots of customer feedback data, using generative AI was an obvious choice for Yabble. They added the following new updates. Tool that analyses customer comments and categorizes them by sentiment and topics. Used GPT-3, to translate data sets into meaningful themes saving manual work for Yabble team. Startups both small and big are at the forefront of accessing AI and solving real business problems. If you are a product developer at a tech company or a startup here are some takeaways from these examples. Think of Gen AI as adding a new intelligence layer to your product. Current Gen AI models are really good at manipulating text. Text summarizations, text categorization, extracting sentiment from a given text are easy but high ROI use cases. Anywhere there is text in your product you can use gen AI to save time for your users. That’s it for Day 15 of 100 Days of AI. I write a newsletter called Above Average where I talk about the second order insights behind everything that is happening in big tech. If you are in tech and don’t want to be average, . subscribe to it Follow me on , or for latest updates on 100 days of AI. If you are in tech you might be interested in joining my community of tech professionals . Twitter LinkedIn HackerNoon here

100 Days of AI, Day 15: How Startups Leverage the Power of Generative AI in Their Products

AI এর 100 দিন, দিন 15: কিভাবে স্টার্টআপগুলি তাদের পণ্যগুলিতে জেনারেটিভ AI-এর শক্তিকে কাজে লাগায়

100 Tage KI, Tag 15: Wie Startups die Leistungsfähigkeit generativer KI in ihren Produkten nutzen

100 días de IA, día 15: Cómo las empresas emergentes aprovechan el poder de la IA generativa en sus productos

100 jours d'IA, jour 15 : comment les startups exploitent la puissance de l'IA générative dans leurs produits

एआई के 100 दिन, दिन 15: स्टार्टअप अपने उत्पादों में जेनरेटिव एआई की शक्ति का लाभ कैसे उठाते हैं

AI の 100 日、15 日目: スタートアップ企業が自社の製品で生成 AI の力をどのように活用するか

AI 100일, 15일차: 스타트업이 제품에 생성 AI의 힘을 활용하는 방법

100 dias de IA, dia 15: como as startups aproveitam o poder da IA generativa em seus produtos

100 дней искусственного интеллекта, день 15: как стартапы используют возможности генеративного искусственного интеллекта в своих продуктах

100 Gün Yapay Zeka, 15. Gün: Startup'lar Ürünlerinde Üretken Yapay Zekanın Gücünden Nasıl Yararlanıyor?

100 ngày của AI, Ngày 15: Cách các công ty khởi nghiệp tận dụng sức mạnh của AI sáng tạo trong sản phẩm của họ

人工智能 100 天，第 15 天：初创公司如何在其产品中利用生成式人工智能的力量

Look, I don't know where you fall on the spectrum of "does emulation promote piracy?", but the idea that a video game company like Nintendo can let off its army of lawyers on a bunch of hobbyists/programers/geeks like a pack of hungry wolves and force them to shut down AND pay over 2 million in dollars just because thinks emulation promotes piracy should alarm any and every one in the tech space. it To quote Steam Founder Gabe Newell: "Piracy is not a pricing issue. It’s a service issue." Amen to that, brother, AMEN! Microsoft's Xbox understands this. Sony's PlayStation understands this. But somehow, the memo was lost on Nintendo which thought it was absolutely ok to force Tropic Haze, the company behind Switch emulator Yuzu and 3ds emulator Citra, to shut down its websites, take down its code repositories AND pay $2.4 million because, apparently, its software allowed people to illegally play Zelda. Nevermind the fact that thousands of people legally pay for Nintendo titles and emulate them on a different hardware because it plays better or because it provides more accessibility. But of course, that's the kind of rhetoric that would fall on deaf ears at Nintendo HQ. Instead, you have Nintendo coming out of the woods to protect its iNtElLeCtUaL pRoPeRtY whenever someone makes a fan project, a ROM website full of abandonware titles, or literally dozens of instances where people just want to do cool and different things with its video game titles. Ironically, except for a handful of titles that Nintendo offers through its online subscription service, there is no "legal" way to play games from the yesteryears without actually emulating them. Emulating video games also ties into video game preservation, but that's a whole different topic altogether. On top of that, Nintendo won't even port its major titles to the PC. Perhaps because it doesn't want to share revenue with others OR it wants to keep selling the Switch consoles. Whatever the case may be, it still SUCKS that Nintendo would just blatantly shut down a project like Yuzu. In some ways, one could argue that this is an attempt by one of the most successful companies in the world to make an example out of a project that is born out of passion. But what does that get the company ultimately? It's already massively rich, and all it's achieving out of these intimidation tactics is to scare off gamers and those who create emulators for them. There's no telling where the scales will tip for emulators for older Nintendo consoles like the Wii or the Wii U, but it wouldn't be surprising if more and more developers decide to not create emulator software because they don't want to fall foul of Nintendo's legal antics. So, yeah. Nintendo sucks and this is the world we live in. 👋 You’re reading HackerNoon's , a weekly collection of tech goodness that combines HackerNoon's proprietary data with internet trends to determine which companies are rising and falling in the public consciousness. Subscribe to receive the newsletter in your inbox every Tuesday. Tech Company News Brief here complete Biden Administration to Dole Out Billions in Federal Grants to Chipmakers🤑🤑 The who's who of the semiconductor manufacturing industry have put a gun to the US government's head and are trying to milk billions of taxpayer money to build their projects on American soil. OK, that's a bit of an exaggeration. The truth is that the Biden administration , REALLY wants companies like Taiwan Semiconductor Manufacturing Company, Intel and Micron Technology to build their products in the US, so much so that it's willing to throw billions of dollars at them. really How much, exactly? Well, the US government plans to dish out over $28 billion in federal grants to each of the companies plus Samsung Electronics to entice them to boost semiconductor manufacturing in the United States. Taiwan Semiconductor Manufacturing is expected to win more than $5 billion in federal grants to support the development of two fabrication facilities in Arizona. Topline numbers for the other companies continue to fluctuate but an announcement for the major advanced chipmakers is expected by the end of this month. Keep your eyes peeled for that development. 🤖 Elon Musk Takes Exception to OpenAI's Company Charter Wow, it's been quite a while since we reported on Elon Musk. But we're happy to let you know that the world's richest man is still up and about with his antics and this time, it involves his old nemesis Sam Altman's OpenAI. Yes, everyone's favorite billionaire has decided to hold OpenAI's company charter to court. He’s arguing in a lawsuit that the (former?) non-profit company, of which he apparently was a co-founder, abandoned its charge to benefit humanity by developing powerful artificial intelligence. Instead, the company rolled out generative AI products like ChatGPT that were essentially Microsoft products to make big bucks. Of course, Musk knows about benefiting humanity given that he owns Tesla; in fact, so concerned was he with humanity's benefit that he decided to burn his own money to buy X (formerly known as Twitter) for reasons that are not yet clear (but should be in a decade or two). all OpenAI is certainly not having Musk's BS, so it has hired a law firm that Elon Musk hates, to have his lawsuit thrown out. really Just another day of tech bros duking it out against each other. Sigh! In Other News.. 📰 Ethereum Finalizes 'Dencun' Upgrade, in Landmark Move to Reduce Data Fees — via CoinDesk Startups are hiring fewer workers and paying out less in equity comp — via TechCrunch EU approves landmark AI law, leapfrogging US to regulate critical but worrying new technology — via CNN US House passes bill to force ByteDance to divest TikTok or face ban — via Reuters The road map to AI's next level could be nature — via Axios Palantir CEO says short sellers bet against great companies ‘so they can pay for their coke’ — via CNBC – And that's a wrap! Don't forget to share this newsletter with your family and friends! See y'all next week. PEACE! ☮️ — Sheharyar Khan, Editor, Business Tech @ HackerNoon *All rankings are current as of Monday. To see how the rankings have changed, please visit HackerNoon's page. Tech Company Rankings

I Hate Nintendo

আমি নিন্টেন্ডোকে ঘৃণা করি

Ich hasse Nintendo

odio nintendo

Je déteste Nintendo

मुझे निनटेंडो से नफरत है

私は任天堂が嫌いです

나는 닌텐도를 싫어한다

Eu odeio Nintendo

Я ненавижу Нинтендо

Nintendo'dan Nefret Ediyorum

Tôi ghét Nintendo

我讨厌任天堂

Get ready for an AI earthquake! A team of UCLA researchers ( n, , , ) have dropped some major keys to AGI. It's not only the code to seriously human-sounding AI, but they've also gone and open-sourced the whole thing. @zxche @Yihe__Deng @HuizhuoY @Kaixuan_Ji_19, @QuanquanGu Now you can develop better LLMs without needing to feed it tons of new, human-annotated data. First, let's focus on the game-changer here: a self-teaching language model. This method lets a language model becoming better and better without massive amounts of new, externally curated data. teach itself, Introducing SPIN: Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models I went full deep-dive mode – read their paper (" "),  scoured the insights on forums like , , and with Google Gemini Ultra and GPT-4 Turbo – and the core concept of SPIN knocked my tech-loving metaphorical socks off: Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models HackerNews X Reddit The 'Conversation Partner' Trick Imagine starting with a language model that has mastered basic skills (let's say conversational etiquette). With SPIN, the model building a dataset from what it already knows. generates internal 'conversations,' Instant knowledge expansion! Step two involves unleashing a new model and giving it one task: spot the difference between and genuine human communication. This forces the original model to up its game, getting more and more with every response to avoid detection. machine-generated chats human-like Here's where things get interesting. They started with (already fine-tuned with ). SPIN unleashed an iterative training system with this base model, improving it exponentially without relying on tons of new externally created data. zephyr-7b-sft-full UltraChat corpus SPIN vs. Traditional AI Training (DPO): A New Champion? We usually think machine learning, particularly for these huge language models, requires boatloads of carefully curated and labeled data. methods involve humans painstakingly rating AI responses against each other for training. Not only is this labor-intensive, but it also balloons costs as a dataset grows. Direct Preference Optimization (DPO) Direct Preference Optimization (DTO) is a training method where a model is fine-tuned using a dataset of preferences, often involving human judgments that decide which of the model-generated responses are preferred. This method requires collecting new data where each piece is labeled based on these preferences, which can be resource-intensive. In contrast, SPIN utilizes , significantly reducing the need for new data. iterative self-play By the first iteration, , highlighting its efficiency and effectiveness in leveraging existing data to enhance model performance. SPIN's performance already exceeds that of DPO in most cases SPIN showcases its strength by achieving . The process of , methodically enhances the model's performance across multiple iterations, , especially on challenging benchmarks like TruthfulQA and GSM8k. on-par performance with models trained on more extensive datasets iterative training showcasing substantial improvements So, SPIN outperforms conventional training methods, including DPO, by efficiently leveraging synthetic datasets generated through self-play, without the need for additional human-annotated data. What are SPIN's Strengths and Costs? SPIN throws a curveball with its self-play dynamic. Think of it like a language model sparring with itself in a linguistic boxing ring, with each round teaching it new tricks. SPIN's data efficiency bypasses the need for new human-annotated datasets. But more importantly, it , . accelerates the improvement loop making the model increasingly adept at generating human-like text Not only does SPIN seem to match models trained on larger external datasets, but its iterative power means consistent gains as it essentially studies its own output. Mindblowing, right? Okay, Let's Talk the Elephant in the Room – COST Nous Research co-founder has a point. These big ol' language models don't get smarter for free. Iteratively re-training with SPIN involves the expensive process of Supervised Fine-Tuning (SFT) each time. @Teknium1 However, he also mentions that "I think its worth it!". Also, the long-term benefits of quicker evolution and potentially less dependency on human-annotated data outweigh the initial investment? That's the exciting question! BOOM! It's Open-Source AI Time Just yesterday, , associate professor of computer science at UCLA and director of AI research at ByteDance, announced that . This doesn't just mean code and datasets, but pre-trained models to kickstart your own AI journeys. Quanquan Gu anyone can now use the SPIN model and dataset SPIN mirrors human thought processes. By generating text that feels human, SPIN hints at the foundational elements of reasoning that future AI could do. You know how some LLM outputs feel robotic right? Well, SPIN is different. It actually mirrors the way humans think. The way it writes feels so natural, it's like a peek into how future AI might be able to reason for themselves. This isn't just about making chatbots sound nicer. It's about creating a kind of digital thinking that works like ours. That kind of AI would be so much more flexible and capable of real understanding. While SPIN is a big leap forward in making language models sound more natural, . it's easy to get excited and overestimate what it means The text it produces is impressive (you can take a look to the database), but it's important to remember that AI doesn't yet have the capacity for true independent reasoning. While SPIN isn't true , the way it mimics human-like writing demonstrates impressive advances in how AI could process and use language in the future. AGI Even so, it does suggest amazing possibilities for how AI and language might develop in the future (if you remember that we are at the beginning of the hockey stick, the future is not far from today...) The ripple effects will be huge and here's your access pass: Code: Available on : GitHub https://github.com/uclaml/SPIN Data: Hosted on , the dataset is readily accessible for those eager to apply SPIN methodologies: Hugging Face https://huggingface.co/collections/UCLA-AGI/datasets-spin-65c3624e98d4b589bbc76f3a… Models: Pre-trained models are also available, offering a head start for experimenting with SPIN-enhanced language models: https://huggingface.co/collections/UCLA-AGI/zephyr-7b-sft-full-spin-65c361dfca65637272a02c40… Project Page: For comprehensive insights and further information, the project page is an invaluable resource: https://uclaml.github.io/SPIN/ To sum up, its iterative, self-improving methodology is a significant advancement towards creating LLM that can engage in genuinely human-like communication. Originally shared on my . X account

AGI Blueprint? UCLA Researchers Open-Source SPIN—A Self-Improving Language Model

এজিআই ব্লুপ্রিন্ট? UCLA গবেষকরা ওপেন-সোর্স স্পিন—একটি স্ব-উন্নতিশীল ভাষা মডেল

AGI-Blaupause? UCLA-Forscher Open-Source-SPIN – ein sich selbst verbesserndes Sprachmodell

¿Plan AGI? SPIN de código abierto de investigadores de UCLA: un modelo de lenguaje que se mejora a sí mismo

Plan AGI ? SPIN open source des chercheurs de l'UCLA : un modèle linguistique auto-améliorable

एजीआई ब्लूप्रिंट? यूसीएलए शोधकर्ता ओपन-सोर्स स्पिन-एक स्व-सुधार भाषा मॉडल

AGIブループリント? UCLA の研究者がオープンソースの SPIN — 自己改善型言語モデルを公開

AGI 청사진? UCLA 연구원 오픈 소스 SPIN - 자체 개선 언어 모델

Projeto AGI? SPIN de código aberto para pesquisadores da UCLA - um modelo de linguagem que se aprimora automaticamente

Проект AGI? Исследователи Калифорнийского университета в Лос-Анджелесе SPIN с открытым исходным кодом — самоусовершенствующаяся языковая модель

AGI Planı? UCLA Araştırmacıları Açık Kaynak SPIN—Kendini Geliştiren Bir Dil Modeli

Bản thiết kế AGI? Các nhà nghiên cứu của UCLA SPIN mã nguồn mở—Mô hình ngôn ngữ tự cải thiện

AGI 蓝图？加州大学洛杉矶分校研究人员开源 SPIN——一种自我改进的语言模型

Any views expressed in the below are the personal views of the author and should not form the basis for making investment decisions, nor be construed as a recommendation or advice to engage in investment transactions. I recently read a fascinating book about the history of the East India Company (EIC) called “The Anarchy” by Wiliam Dalrymple. For those unfamiliar with this chapter of European colonialism, the EIC was a joint stock company that was given a royal charter/monopoly over the trade between Britain and the Indian subcontinent. Over a few centuries, what was initially a weak and poor commercial enterprise that survived at the whim of various Indian rulers conquered the entire subcontinent and paved the way for the British Raj, which lasted from the late 19th century until 1947. When it became known that the EIC employed unsavory methods to extract its immense profits, certain members of the company were called in front of the British Parliament for questioning. Fortunately, because many high-ranking members of parliament were also EIC shareholders, scant, if any, punishments were ever handed out. In fact, on various occasions, the EIC requested and received government bailouts due to its over-indebtedness. EIC was the first Too Big to Fail company, and conflicted politicians back then did as their contemporaries do today … use the public purse to bail out private companies for their own personal benefit. In effect, privatize the gains, but socialize the losses. The EIC is apropos to crypto because I want to talk about the evolution of crypto project ownership structures and fundraising methods. I will write about Bitcoin, ICO’s, yield farming/liquidity mining, and finally, points. The goal of this essay is to provide context for why points as a way of engaging users is a natural progression that follows on from previous engagement and fundraising methods of the past. Many of Maelstrom’s portfolio companies will launch their tokens in 2024, and you can expect to learn about points programs, which all share a goal of generating usage of their protocols. Therefore, I want to discuss why points exist and how they will drive adoption in this cycle. Web 2 vs. Web 3 Who your shareholders/token holders are and what they are promised is very important to any commercial enterprise's success, whether in crypto or otherwise. I am specifically going to compare and contrast how Web 2 (technology startups) and Web 3 (crypto startups) fundraise and acquire users. Whether you are founding a Web 2 or Web 3 startup, acquiring and retaining users is the most challenging and expensive aspect of your business. In Web 2, which was most prevalent from 2010 to 2020, a VC fund looked for startups with a modicum of initial traction and subsequently provided cash money rocket fuel for the startup to go on a user acquisition spree. Usually that entails handing out the service for free or at a discounted rate, well below its actual cost of delivery. Remember when ride-sharing apps were all fighting viciously for market share, and fares were incredibly cheap? That was all paid for with billions of dollars worth of VC money. Think of it as a subsidy in return for users. At the end of the rainbow for a VC firm was a successful initial public offering (IPO). The IPO allowed the plebes to own a piece of a successful Web 2 company for the first time. IPOs are the way to dump on retail in TradFi. However, various regulations barred the plebes from crowdfunding early-stage Web 2 companies. The irony is that the vast number of plebe users who drove the success of the company were prohibited from owning a piece of it. Considering the law of averages, more than 90% of all new companies fail. Investing in early-stage Web 2 companies is a recipe for losing all of your money. Even though that is the case, most investors, in a normal Gaussian distribution sense, believe they are 3-sigma traders even though they always seem to earn returns around the mean. What this means is that the plebes always want in early on the sure thing ex-post but throw a fit when they invariably lose money on the myriad of failures. At that point, the plebes point the finger at the government because, in our modern age, the average person believes it's the government's job to live their lives for them. I don’t blame them for that belief; the politicians all outdo themselves to paint the picture that nothing in life can ever go wrong if you support them. From a securities regulator’s perspective, there is no upside to allowing poor people to crowdfund early-stage companies. For every Facebook, there are 1,000 MySpaces. Why lose the shot at a promotion because you got sacked for allowing a gaggle of plebes to blow their paycheck on some piece-of-shit company they got sold by a bucket shop huckster? That’s my regulator-sympathetic reasoning behind why plebes aren’t allowed to collectively invest in early-stage companies. My more cynical explanation is that the TradFi gatekeepers earn a fuck ton of fees from IPOs. Let’s go down the list of who gets paid from the IPO process: The investment banks that underwrite the IPO syphon off between 2% to 7% of the capital raised. The lawyers make hundreds of thousands and sometimes millions of dollars preparing and filing the prospectus and other offering documents. Being a paper pusher never paid so well. The audit and accounting firms make hundreds of thousands of dollars per deal creating the audited financial reports. Quickbooks on steroids! The exchanges charge hefty listing fees. Nasdaq Labs anyone? The cartel of trust described above loves IPOs. A few successful IPOs in a calendar year ensures that everyone in TradFi gets their phat bonuses. But without a hungry hoard of retail buyers who are excluded from investing earlier at much cheaper prices, there would be no buying pressure to create a successful IPO. That is why retail participation must be saved for the end and not the beginning of the fundraising lifecycle. While these fees may seem egregious, before Bitcoin, there was no other efficient way to raise funds publicly. And being objective, this process created many super important, useful, and profitable companies. It worked. But fuck nostalgia, let’s move forward. From a user’s perspective, the major issue with how Web 2 companies are formed is that using the product or service doesn’t accrue equity in the company providing it. You don’t get shares of Meta by ogling Instagram thirst traps. Nor do you get shares in ByteDance while you destroy your brain watching pre-teens dance like Cardi B on TikTok. These centralized companies take your attention, your actions, and your money, and yes, they provide a service or product you enjoy, but that’s it. And even if you want to invest, you can’t unless you are rich and well-connected. Participation ≠ Ownership Bitcoin Paradigm Change Bitcoin and the subsequent evolution of the crypto capital markets changed this. As of 2009 – the genesis Bitcoin block – it became possible to reward participants with ownership in the startup. This is what I will refer to as Web 3 startups. Before you could purchase Bitcoin on an exchange starting in 2010, mining it was the only way to acquire it. Miners, by burning electricity, validate transitions, which creates and upkeeps the network. For this activity, they are rewarded with newly minted Bitcoin. Participation = Ownership The participants or users now have skin in the game. Similar to the EIC example, users with skin in the game will do whatever they can to protect their investment. In the EIC case, that meant looking the other way as war and famine were visited upon a population in the name of profit. In the Bitcoin example, it means owners of Bitcoin want to convert as many people as possible into Bitcoin users as well. Look into my laser eyes, bitch! The Initial Coin Offering (ICO) Quickly, the crypto capital markets recognized a way to crowdfund technology startups in a way that was unavailable to Web 2 startups. If a Web 3 startup eschewed fiat as payment for ownership or governance rights and only accepted Bitcoin, it could sidestep the entire TradFi cartel of trust that “safeguards” the filthy fiat financial system. To raise funds, a project launched a website announcing that if you gave them Bitcoin, they would give you tokens that did something or gave some sort of economic rights in the future. If you believed this team had the ability to execute their advertised vision, in less than an hour, you could own a piece of an exciting new network. The first major project that conducted an ICO was Ethereum in 2014. The Ethereum Foundation pre-sold Ether, the commodity that powered its virtual decentralized computer, in exchange for Bitcoin. In 2015, Ether was distributed to buyers. It was a successful ICO. The foundation pre-sold its token for funds that were used to develop the network further. The amount of ICO deals and the amounts raised went asymptotic. The best ICO ever done was by Block.one, the creators of EOS. They conducted an ICO over a whole year and raised $4 billion worth of Ether at prices back then. EOS is an excellent example of the ICO craze because it raised the most money, and it was a complete piece-of-shit blockchain that barely worked. As time went on, the projects got shittier, but the amounts raised grew larger. This was because for the first time ever, anyone with an internet connection and some crypto could own a piece of what were sold as the next hottest technology startups. Retail entered the game and made a lot of crypto bros filthy rich. The ICO mania peaked in the fall of 2017 when the Chinese regulators explicitly banned them. Exchanges like Yunbi were shut overnight, and many projects that recently raised money from Chinese retail investors returned the funds. Around the world, regulators keen to protect the cartel of trust from competition in how early-stage ideas are fundraised swung into action, and ICO issuance disappeared. ICO investors gave money and became motivated marketing agents for the projects. However, just because you sold tokens at a high price didn’t mean anyone would use your product. Ownership was just a function of money invested but not usage of the protocol. ICOs were a step forward, but we could do better. Yield Farming DeFi summer, from a northern-hemispheric perspective, started in mid-2020. A slew of projects launched during the depths of the 2018-2020 bear market started being used meaningfully. The tokens were already listed as many of these projects had either done an ICO or a token pre-sale followed by a public token generation event (TGE) by mid-2020. The project foundations allocated a large amount of tokens to be awarded to community members who performed valuable activities. Many of these projects, like Uniswap, AAVE, and Compound, were focused on borrowing, lending, and trading. They wanted users to borrow, lend, and/or trade their crypto assets using their protocols. In return for doing said actions, the protocols would instantly emit freely tradable tokens. And thus, yield farming was born. Traders borrowed, lent, and traded cryptos on these platforms for the specific purpose of earning the governance token of the protocol. In many cases, the trader would lose money just so they could “farm” or earn more tokens. Because the tokens went up in price on a mark-to-market basis, the traders appeared to be in profit. All of this assumes you sold at the top; many didn’t, and all that activity was for naught. From the project’s perspective, all this activity juiced their trading volumes, total value locked (TVL), and number of unique wallets that interacted with the protocol. These metrics convinced investors that DeFi was working and creating a parallel financial system governed purely by code run on decentralized virtual computers. Participation = Ownership All was good, except the unlocked token supply grew too quickly for many projects. Invariably, the projects had to slow emissions, and the market started to ask, “What’s next?” If all activity is predicated on aggressive token emissions schedules, what happens if the token price falls and there are no more tokens available to give prospective users? What happens is the price of the project’s token plummets alongside activity. The lesson learned is that yield farming or liquidity mining is a great way to incentivize usage, but if it is pursued too aggressively it becomes a liability. The question then becomes, how do you emit tokens in a more sustainable manner? Points Yield farming as a user acquisition tool died alongside the 2021 crypto bull market. But points were created in its wake, and have quickly become the go-to pseudo-ICO fundraising and user acquisition tool for projects in this current bull cycle. Points combine the best aspects of ICOs and yield farming. ICOs Allow the millions of retail crypto hodlers to purchase a piece of a new protocol. But when you sell something to retail, some regulators call that a “security” and require you to do a lot of shit that you don’t want to do … namely, stop selling shit to poor people. Yield Farming Emits tokens to users for using the protocol. However, if pursued too aggressively, it will inflate the finite token supply too quickly, and once the token price falls, the user will have no incentive to use the protocol any longer. What if a project gave you points for interacting with the protocol? These points would be converted into tokens, which would then be airdropped free of charge into users' wallets. What if the points to airdrop token price was completely opaque and at the discretion of the project? What if there was actually no promise that points would even convert into future airdropped tokens? Here is a simple example to bring it home. Suppose that Sam Bankman-Fried (SBF) is offered a points program by his bunkmate. His bunkmate can get an infinite supply of Emsam; this is the amphetamine SBF claims he needs for his ADHD condition. His bunkmate really likes back massages. He makes a deal with SBF. For every back massage SBF administers, his bunkmate will give him a point. At a future date of his bunkmate’s choosing, the points might convert into a certain number of Emsam pills. SBF wants his meds so bad he is willing to agree to a lot of rubbing in exchange for a soft promise of future Emsam pills. Are points a contract between the project and the user for a tangible reward in the future? No. Is there any form of money, fiat, crypto, or otherwise that is exchanged between the user and the project for points or tokens? No. Does the project have complete flexibility regarding the points-to-token conversion price and the timing of a token airdrop, if one happens at all? Yes. Let’s make some other assumptions and observations. If you have a small, talented engineering team, there is no need for many other employees. That is the beauty of software. The layer-one blockchain itself handles cybersecurity; users pay gas in the native token, such as Ether, in part to pay validators or miners to secure the network. Do you need in-house lawyers? If your project is truly decentralized, maybe not. And furthermore, after your foundation is created, what other major work is there for expensive lawyers? Your biggest problem is how to get more users and that is a marketing and business development effort. Assuming you built good tech, all your money will be spent doing things to attract usage. Building an amazing project without little, if any, VC funding is entirely possible. The project needs money to acquire users, and points are the new and exciting way to guerilla market. With points, the project doesn’t lock itself into an aggressive token emissions schedule. That is because the points-to-token ratio can be changed at any time. There is no contract that says the project must match a specific points-to-token ratio. With points, the project generates usage in the specific ways it believes will increase the long-term value of its service. Many of the most impactful crypto projects are some sort of two-way marketplace. Points help jumpstart network activity and overcome the cold-start or chicken-and-egg problem. The ability to surgically, dynamically calibrate points emissions to specific actions throughout the project’s ecosystem means that the project can be very effective in generating the exact type of user interaction it desires. Finally, with points, the project doesn’t have to lean as heavily on signing token pre-sale deals with VCs and other high-net-worth investors. The major purpose of VC capital is to pay for user acquisition; points can accomplish this. The best part is that the project can implicitly sell its token at a much higher price via the opaque points program than a transparently priced round. Points are great for the project, but what about retail users? Since ICOs went out of favor, retail must be super wary of VC unlock schedules. If retail apes into a token at the same time a VC’s portion unlocks, it’s REDRUM REDRUM REDRUM for retail’s portfolios. Using points, projects don’t need to engage in pre-sale token sales to such a large extent. Using points, retail can “invest” earlier and hopefully get a cheaper price than if they waited until after the TGE. While ambiguity surrounds the timing of the airdrop and the points-to-token ratio, points might represent a more equitable way to reward users for their participation. A points program is only effective if there is a high degree of trust between users and the project’s founders. The user trusts that after interacting with the protocol, their points will convert into tokens at a reasonable price in a reasonable time frame. As points programs proliferate, there will be bad actors who abuse this trust. Ultimately, an egregious breach of trust that encompasses a large amount of money might be what kills points as a fundraising and user engagement tool. But we aren’t there yet, so I ain’t fussed. All Aboard Whether you like it or not, every successful project, and by success I mean, token number go up, will enact a points program before their TGE. This will create usage of the protocol, hype around the probable token airdrop, and a Pump Up The Jam! public listing. I am a businessman, not a priest. The only dogma I subscribe to is “number go up!” If points create better alignment between users and the protocols, LFG. If the cartel of trust’s hold on fundraising of the new and innovative technology startups further erodes because points are considered a better user engagement and fundraising mechanism, sign me up. I hope this essay provides some context on what points are and why I believe they will power this cycle's best-performing token launches. Maelstrom’s got bags, and I’m not ashamed to tell readers about them. Expect many more essays on exciting projects in our portfolio that launch points programs before their eventual TGE. My body is ready, is yours?

ByteSync

bytesyncapp.com

1-10 employees

Since n.d.

COMPANY RANKING

BYTESYNC

EVERGREEN INDEX #2541

ByteSync WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!

ByteSync

bytesyncapp.com

1-10 employees

Since n.d.

COMPANY RANKING

BYTESYNC

EVERGREEN INDEX #2541

RELATED COMPANIESRANK

STORIES ABOUT BYTESYNC

ByteSync WIKI

Apply to get your Tech Company News Page live on HackerNoon.com today!