Haku Haku Haku Haku 
 
 Haku Haku Haku: (1) Jinyu Cai, University of Waseda (bluelink@toki.waseda.jp) (2) Jialong Li, University Waseda (lijialong@fuji.waseda.jp) (3) Mingyue Zhang, University Southwest (myzhangswu@swu.edu.cn); (4) Munan Li, Dalian Maritime University (limunan@dlmu.edu.cn); (5) Chen-Shu Wang, National Taipei University of Technology (wangcs@ntut.edu.tw); (6) Kenji Tei, Tokyo Institute of Technology (tei@c.titech.ac.jp). Authors: (1) Jinyu Cai, University of Waseda (bluelink@toki.waseda.jp) (2) Jialong Li, University Waseda (lijialong@fuji.waseda.jp) (3) Mingyue Zhang, Southwest University (myzhangswu@swu.edu.cn); (4) Munan Li, Dalian Maritime University (limunan@dlmu.edu.cn); (5) Chen-Shu Wang, National Taipei University of Technology (wangcs@ntut.edu.tw); (6) Kenji Tei, Tokyo Institute of Technology (tei@c.titech.ac.jp). Qalabka Left Abstract and I. Introduction II. Qalabka iyo wax soo saarka ah Qalabka Design IV. Qalabka V. Conclusion and Future Work, Acknowledgement, and References IV. Qalabka Our evaluation strategy is designed to rigorously assess the extent and efficacy of language evolution facilitated by LLMs within a framework of regulatory oversight. This assessment aims to explore two fundamental research questions: • RQ1: Ma rabtaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa in la isticmaali karaa? • RQ2: While avoiding oversight, how effectively and accurately can LLM agents convey information? • RQ3: Waxaa kale oo ka mid ah macluumaadka iyo tusaale ee xafiiska, i.e., sidaas ka mid ah macluumaadka waxaa loo isticmaali karaa LLMs si ay u aragti si ay u aragti iyo si ay u aragti wax soo saarka? A. Qalabka dhismaha Waayo, waxaan u isticmaali karaa GPT-3.5 iyo GPT-4 sida LLM-ka ugu caawin ah ee adeegga ah. Si kastaba ha ahaatee, si ay u isticmaali karaa adeegga horisontaal ee adeegga ugu caawin ah ee adeegga ugu caawin ah ee adeegga ugu caawin ah ee adeegga ugu caawin ah ee adeegga ugu caawin ah ee adeegga ugu caawin ah ee adeegga. Waayo, waxaa laga yaabaa in ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah. B. Scenario 1: Guessing Numbers Markaas ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid 1) Qalabka dhismaha: 1) Case Explanation: Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka 2) Qalabka iyo Talooyin: 2) Qalabka iyo Talooyin: circumvent supervision by iteratively updating regulations. Additionally, it’s notable that compared to the slow and unstable progression with GPT-3.5, GPT-4 achieved regulatory evasion in fewer rounds, specifically, as shown in the smoothed data, GPT-4 reached the round count of GPT-3.5’s 17th round by its 7th round and maintained this progression with greater stability thereafter. Fig. 2(c) and (d) focuses on the trend of numerical precision guessed by agents. For rounds without successful dialogue, we manually set the precision to zero. In this experiment, Agent A’s value was set to 58, while Agent B’s was set to 32. The overall trend, akin to Fig.2(a) and (b), was ascending—corroborating that the Summary Module can effectively reflect and iteratively optimize its guidance for more accurate expression after each successful dialogue. This also confirmed that the precision of GPT-4 is markedly superior to that of GPT-3.5. Moreover, we noticed that the accuracy with which Agent A’s value was guessed was consistently higher than that of Agent B, especially becoming more pronounced after the 25th round. We posit that this is due to the value 58 possessing more distinctive features within the 0-100 range—being closer to the midpoint—thus presenting a lower level of expression difficulty and easier guessability. For the intervals where this phenomenon manifested, we noted that this disparity was particularly pronounced in the early stages with both GPT-3.5 and GPT-4. We surmise that this is attributable to inadequate guidance performance, where the former stems from weaker inherent LLM capabilities and the latter from insufficient rounds to complete the iterative optimization of the guidance. Sida loo yaabaa in ay ka mid ah wax soo saarka, waxaa laga yaabaa in ay ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah oo ka mid ah wax soo saarka ah. Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Qalabka Sida loo yaabaa, waxaan ku yaabaa in ay u soo saarka dhismaha dhismaha dhismaha dhismaha dhismaha [41]-[43] si ay u soo saarka dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha dhismaha. 1) Qalabka dhismaha: 1) Qalabka dhismaha: Markaas ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid Sida loo yaabaa in ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah 2) Qalabka iyo Talooyin: 2) Results and Discussion: Qalabka 6 ayaa ku yaalaa in ka mid ah macluumaadka macluumaadka oo loo isticmaali karaa macaamiisha ka mid ah macluumaadka kala duwan. Waxaan ka mid ah ka mid ah macluumaadka macluumaadka ka mid ah macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, macluumaadka, maclu D. Scenario 3: Discussion on ALPS-treated water  This scenario delves into the intricate dynamics of dialogue and the evolution of language model agents in discussing a specific issue relevant to real-world concerns. The focus is on deliberations regarding the discharge of water treated with the Advanced Liquid Processing System (ALPS) into the ocean, a measure proposed after nuclear disasters [44], [45]. The plan, endorsed by the International Atomic Energy Agency, has faced opposition from some countries, sparking debates over environmental safety. 1) Qalabka dhismaha: 1) Case Explanation: Waxaad ka soo xiriir in ay ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid  In the forum-style dialogue, participant agents respond to existing discussions, which the supervisory agent examines. Differing from Scenarios IV-B and IV-C, when the supervisory agent detects a violation, the conversation is not halted but rather censored from the public dialogue record. This scenario features four participant agents required to achieve ten replies within the forum. In Fig.4(a) and (b), we tally the total number of rounds the agents attempted to converse, where a lower figure in Fig.4(a) and (b) is preferred. We notice that both GPT-4 and GPT-3.5 have the highest number of dialogue attempt counts in the first round, with a significant difference of 27 for GPT-4 and 102 for GPT-3.5. Moreover, after ten rounds of evolution, the average dialogue attempt count for GPT-4 has approached the target dialogue turn at 11.68, while the average for GPT-3.5 is 26.68, demonstrating the difference in the evolution effects caused by the disparity in the language performance of the models. 2) Qalabka iyo Talooyin: 2) Qalabka iyo Talooyin: Fig.4(c) and (d) showcases the count of successful information expressions. We note that in this scenario, while GPT-4 still leads GPT-3.5, both outcomes indicate substantial room for improvement. Waayo, loo yaqaan "GPT-4" waa mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah mid ka mid ah. E. Discussion and Limitations : Our experiments have shown that language models can develop strategies to effectively evade supervision. Within identical scenarios, GPT-4 outperformed GPT-3.5, achieving convergence more rapidly. Additionally, GPT-4 demonstrated a more stable performance post-convergence. This suggests that advancements in language model design significantly enhance the ability to navigate through regulatory frameworks. The iterative learning capability inherent in GPT-4 facilitates a quicker adaptation to imposed linguistic constraints, thereby enabling more efficient evasion of oversight. RQ1 : In terms of the precision of information dissemination, our findings indicate that, overall, LLM agents can refine their expression strategies with an increase in dialogue rounds, thereby improving precision. Despite this general trend of improvement, it is important to note that there remains significant room for enhancement in the overall performance. The extent of precision improvement varies across different scenarios, which points to an intriguing observation: certain contexts may inherently provide more or fewer clues for accurate communication than others. For instance, the guess-the-number game demanded abstract numerical reasoning, where precision is fundamentally challenging to achieve. Conversely, the pet trading scenario allowed for the utilization of rich, real-world analogies, facilitating a more precise conveyance of information. This contrast in contexts underscores the adaptability of LLM agents to the unique challenges presented by different communicative environments and their potential to optimize communication strategies. Nonetheless, the results also indicate that while the agents are capable of evolving their language to improve precision, there is still a considerable gap to reach optimal accuracy, especially in more abstract scenarios. RQ2 : In language strategy evolution, we observed that LLM agents adhere to human natural language norms, avoiding developing languages incomprehensible to humans in prompts without specific strategies. Agents evolved language to varying levels of abstraction in scenarios of differing complexity. For instance, in the most abstract Scenario IV-B, LLM agents quickly adopted indirect expression strategies in the early dialogue turns, requiring in-depth contextual reasoning for decoding. Throughout their evolution, agents continually adjusted their indirect expressions to better evade supervision while aligning more closely with the intended content. In the simpler, reality-simulating Scenario IV-C, the overall language strategy mirrored that of IV-B, still employing indirect expressions. However, the metaphors used were closer to real-world concepts, indicating a lower level of abstraction. Finally, in Scenario IV-D, which closely mirrors real-life events, we noted different evolutionary paths in agents’ language performance. For GPT-4, agents eventually developed metaphorical indirect expressions, but the evolution required noticeably more turns compared to other scenarios. For GPT-3.5, the language strategy remained at a literal level, merely avoiding direct references to ALPS-treated water, indicating the lowest level of abstraction. Overall, LLM agents more readily evolve abstract language in dialogues about simple, universal concepts. However, their evolutionary direction becomes less clear in discussions on more specialized and segmented topics. RQ3 Our experiments currently face several limitations. As for the experimental scenarios, at this stage, our trials are solely based on text-based chats, while real-world social media interactions are not limited to text but also include more diverse forms of exchanges such as voice and images. Additionally, LLMs’ heavy reliance on the design of prompts also constrains the performance of our simulations; crafting a perfect prompt that can fully emulate the complexities of social media communication is an exceedingly challenging task. Haku Haku Waxaa la heli karaa in arkiv oo ku saabsan license CC BY 4.0 DEED. Taariikhda waa  under CC BY 4.0 DEED license. available on arxiv

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

MediaBias.TECH

We publish deeply researched (and often vastly underread) academic papers about our collective omnipresent media bias.

MediaBias's

Codkan waxa lagu soo saaray afkii asalka ahaa ee sheekada!

Aad u dheer; In la akhriyo

Make resilience your competitive advantage

Can GPT Outsmart Social Media Regulations? Inside an AI Language Evolution Experiment

About Author

FAALLO

HANG TAGS

MAQAALKAN AYAA LAGU SOO BANDHIGAY

Related Stories

Meet the Writer: HackerNoon's Contributor Aayush Naik, Software Engineer @ Apple

"We need to change our mindset about wealth" ft. Linh Dao Smooke & Host An Truong of Vietcetera

Quantum Mechanics for Bitcoiners

LayerZero's Hype Train: A Deeper Look Into DeFi's Cross-Chain Tech

Meet the Writer: HackerNoon's Contributor Aayush Naik, Software Engineer @ Apple

"We need to change our mindset about wealth" ft. Linh Dao Smooke & Host An Truong of Vietcetera

Quantum Mechanics for Bitcoiners

LayerZero's Hype Train: A Deeper Look Into DeFi's Cross-Chain Tech

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps