Generative artificial intelligence (AI) could radically transform the business landscape, opening new doors to innovation in product design, customer service, medical discoveries, and much more. But the widespread implementation of generative AI remains in a holding pattern.

Bridging the Confidence Gap in Generative AI

Despite its immense potential and widespread hype, machine learning and artificial intelligence analytics do not always live up to expectations when deployed in business. This is often due to factors such as irrelevant or misaligned use cases,

How machine learning can work for business

Debanjan Saha is the CEO of DataRobot, an A.I. unicorn that serves Fortune 500 companies globally. Mr. Saha is a visionary technologist with a track record of building innovative products and multibillion-dollar businesses at Google, AWS, and IBM.

Debanjan Saha

ChatGPT-style AI can tackle the drudge work of responding to RFPs faster than humans. Sales teams at Google, Twilio, and others say productivity is spiking.

Generative AI Is Coming for Sales Execs’ Jobs—and They’re Celebrating

Foote Partners’ Q3 report on IT skills pay trends show AI and security skills were in high demand, and the value of cash-pay premiums was more volatile but their average value across a broad range of IT skills and certifications was slightly down.

Skilled IT pay defined by volatility, security, and AI

DataRobot’s AI platform provides tools for bias and fairness analysis. Image: DataRobot Beyond adapting established MLOps strategies, Generative AI invites the creation of bespoke tools to meet ...

AI Observability | Monitoring Generative AI's New Capabilities

— Diego Oppenheimer announced he is joining San Francisco-based startup Guardrails AI as co-founder and board member. He previously co-founded Seattle machine learning startup Algorithmia, which was acquired by DataRobot in July 2021. Oppenheimer is partner and CEO-in-residence at venture capital firm and incubator Factory.

Tech Moves: Seattle tech vet joins new AI startup; Rippl, Tasso, Lexion add execs; and more

Before that, he held managerial roles at PricewaterhouseCoopers, and was most recently CFO at DataRobot. Fletcher recruited former Tableau colleague Michael Arvold to be the company’s co-founder ...

Former Tableau execs launch new Seattle startup that helps companies manage cloud costs

Former Tableau CFO Damon Fletcher is coming out of retirement to launch a startup that helps tech companies get a hold of their cloud costs amid a surge in spending driven by artificial intelligence.

That’s where DataRobot Inc. comes in. Gannett is a mass media holding company with names such as the USA Today Network under its umbrella. While gen AI presents vast opportunities in the media ...

How DataRobot facilitates gen AI’s conscientious application: The Gannett perspective

DataRobot

datarobot

AI Isolation Strategies for Businesses

Introduction How many times have you heard the word "AI" in 2023? There is a probability that you must have heard it too many times, especially with the popularity of ChatGPT. AI, which is an acronym for Artificial Intelligence, is a fast-growing phenomenon in today's world. According to statistics, . From education and health to entertainment and business, AI is reshaping our environment and our way of life. the global AI market is valued at over $136 billion and is projected to increase by over 13x over the next seven years At the same time, it is opening up new possibilities and posing new risks to the continued growth, welfare, and survival of mankind. In the modern world, AI is a rapidly expanding phenomenon. Statistics show that the global AI market is already worth over and is expected to grow by more than 13 times over the following seven years. $136 billion AI’s Basic Working Principle… Artificial intelligence is basically the capacity of a digital computer or computer-controlled robot to carry out actions frequently performed by intelligent individuals. Systems and tools that are built with AI can recognize, combine, and deduce information from a variety of data sources, including audio, photos, videos, text, and more, and they can also learn from this information. They can learn from prior experiences, generalize, use reasoning, find meaning, and adapt to new circumstances, which helps them become more intelligent. AI Tools Across 11 Different Sectors in the Modern World Most industries throughout the world—if not all—use AI tools and technology in some capacity. Although it mostly has some beneficial effects, there is concern that given how quickly AI is expanding and improving, and it may have the ability to eliminate many occupations. How accurate is that? But first, let's examine some of the well-known industries that have begun making major use of AI to make this determination. 1. Finance There are powerful AI tools that are used by a variety of financial organizations and even individuals in the areas of financial planning and analysis, assessing credit risk, fraud detection, trading, and investments, optimizing insurance claims processing, sales, and customer services, and digitizing historical files and documents, amongst other applications in the field of finance. Some of the tools include: A virtual assistant tool in the Bank of America mobile app that assists users with a variety of banking operations like checking balances, paying bills, and sending money, as well as balance reconciliation. Capital One also has a similar virtual assistant called  Eno, which notifies users of suspected fraud, price increases, or payment reminders. Erica: Enova uses AI technology to aid users have access to credit facilities and helps financial institutions to solve the challenges surrounding borrowers’ assessment through the provision of sound financial analytics. : Enova An AI-powered platform using data points to provide financial institutions with the data to conduct assessments on borrowers with little or no credit information : Zest : A technology that makes use of artificial intelligence to accurately process financial records like pay stubs, tax returns, and other related ones at a faster speed. Ocrolus provides a predictive model that helps financial institutions solve issues surrounding marketing, wealth management, fraudulent credit card transactions, underwriting, and business growth. : DataRobot an identity verification system that uses ML and AI to analyze and verify applicants’ data to ensure that the KYC details are accurate. : Socure 2. Blockchain Several techniques, including strengthening data security and privacy, improving data quality and trust, optimizing data processing and storage, creating data insights and value, integrating data across platforms and domains, etc., are employed by AI to reinvent and optimize the blockchain sector. Some instances include: - using AI and formal verification to authenticate the security of a blockchain, smart contracts, and web3 dApps. CertiK - which certifies and verifies valuable pieces of art and NFTs on the blockchain using AI. Verisart - which leverages AI to allow quicker transactions, more scalability, and less energy use. Matrix AI - employing artificial intelligence to examine cryptocurrency market movements and offer market insights for individual investors. Token Metrics - which employs AI to build a decentralized network of AI services that can communicate with one another and trade value on the blockchain. SingularityNET 3. Transport AI is revolutionizing transportation and supplying driving assistance. Some of the tools include: This company uses AI to power its autonomous vehicles and taxis for urban transportation. : Waymo Rapid Flow Technologies uses traffic management to deliver adaptive traffic signal control systems that lower congestion and pollution. This is done by analyzing traffic patterns, predicting traffic flow, and optimizing traffic lights using data and algorithms. : Rapid Flow Technologies Using AI, Uber apps are designed to find compatible drivers and passengers, calculate prices and arrival times, and offer safety features. These enhance passenger transportation services by personalizing travel options, streamlining pricing and scheduling, and enhancing customer satisfaction. : Uber Clear uses AI to build biometric identification verification services that let users use their faces or eyes as boarding credentials in airports.  This expedites airport security and boarding procedures in the aviation industry. : Clear With AI, Volocopter has been able to build its drone Taxis which will safely traverse the airspace, avoid obstructions, and land at designated places within a busy urban city. It offers a faster transportation system through congested streets by flying in the lower airspace with zero-emission. : Volocopter 4. Data Analysis One of the key industries using AI and machine learning is data analysis, which does so in a variety of ways, such as by analyzing unstructured data, producing data insights, streamlining data processing, integrating data across platforms and domains, and others. A few of the numerous AI tools that are used in data analysis include: A software program that forecasts data performance using artificial intelligence and suggests subjects based on prediction scores. : Tableau An AI tool designed by Microsoft for Data Analytics, Data Visualization, and Business Intelligence : Power BI A platform to analyze user data and offer insights and recommendations : Research AI An open-source tool to build customized data pipelines, workflow, and solutions for a range of data analysis task : KNIME An AI tool for analyzing and indexing a huge volume of data in real-time : Splunk 5. Software Engineering The coding and design industry is being reinvented and optimized by AI in several ways. Currently, programmers can generate, edit, optimize, check, and transform codes with the use of artificial intelligence (AI). Some of the AI tools available at their disposal include: A tool called , which can create code in common development environments, is , which is based on the GPT-3 standard. GitHub Copilot powered by the OpenAI Codex model A program called ** **utilizes artificial intelligence to examine code and identify flaws that would prevent it from running. Embold This tool is an app that provides programmers with comprehensive facilities to build, edit, and test codes and applications. Examples of popular IDEs include VS code, Eclipse, IntelliJ IDEA, PyCharm, Xcode, Android studio, NetBeans, Atom, Sublime Text, and Vim. The integrated development environment (IDE): An auto-completion tool that can be integrated into IDEs to improve productivity. This tool employs deep learning algorithms to predict the rest of the code a programmer is typing. : Deep TabNine : This AI-powered coding assistant using natural language, can generate codes, complete the function of the codes, and generate the summary of a function. CodeT5 Other AI tools that aid programmers include Google Bard, ChatGPT, AskCodi, Mutable, IBM Watson Code Assistant, Code-LMS, WPCode, and Amazon CrowdWhisperer. 6. Entertainment Although technically speaking, humans make up the majority of the characters, artists, and actors in the entertainment industry, artificial intelligence (AI) has begun to play crucial roles in the sector and is used in significant areas like generating, analyzing, optimizing, moderating, and personalizing realistic content for their audiences by using data and algorithms. Some platforms for entertainment that use AI are: aids artists in writing the lyrics of a song. It can write songs in over 100 languages. Lyrical Labs offers video analytics and insights across a variety of platforms, including YouTube, Facebook, Instagram, and TikTok. Tubular Labs is an AI-powered platform using machine learning algorithms to match licensed music with visual content. SyncFloor This is an AI video creation tool that can convert text data to video content. Lumen5: Other global organizations that inculcated AI into their products are Netflix__,__ which uses AI to adapt its streaming quality and bandwidth use to the customers' devices and networks; YouTube, which employs AI to search and eliminate content that breaks its community standards; and Spotify which utilizes AI for curating customized podcasts and playlists for its subscribers depending on their listening patterns. 7. Manufacturing This industry makes use of AI in several areas, including boosting predictive maintenance, streamlining manufacturing procedures, automating production duties, fostering production innovation, and more. The following are some instances of production platforms using AI in their operations: offers AI-based visual inspection solutions for a range of industries, including textiles, electronics, and the automobile. Landing AI for a variety of industries, including energy, mining, and transportation. Uptake leverages AI to offer predictive maintenance solutions. Uptake __c__an give solutions for the optimization of processes in a variety of industries, including the chemical, steel, and cement industries, by utilizing AI. __ m __ Petuu offers robotic picking solutions for a range of industries, including e-commerce, shipping, and the food industry. Covariant For generative design solutions across a range of industries, including architecture, engineering, and construction, leverages artificial intelligence (AI). Autodesk 8. Gaming AI may be used to create intelligent, human-like NPCs that communicate with players, forecast player behavior to improve game design and testing, provide lifelike images and audio, improve gameplay, and increase immersion, among other things. By using data and algorithms, it is also utilized to create, analyze, optimize, moderate, and personalize realistic material for their games. Several instances include: AI is used by to deliver statistics and insights for games across a range of platforms, including PC, mobile, and console. Unity AI is used by to recommend the best times to release mobile games as well as their prices and marketing techniques. Google Play In multiplayer games, employs AI to detect cheaters and block them. Steam 9. Healthcare AI activities in the healthcare industry are increasing by the day. Using systems like Natural Language Processing, Machine Learning, and deep learning algorithms, its impact is greatly felt, especially in domains such as; illness detection and patient care improvement, forecasting and epidemic prevention, medication discovery and development, personalizing and optimizing therapies, and much more. Some notable instances of AI in healthcare  include: The use of AI by to help oncologists diagnose and treat cancer. IBM Watson Health AI-powered tool to track and predict infectious disease outbreaks globally. BlueDot tool used to organize searches across billions of chemicals for possible drugs to speed up drug development. Atomwise Provision of individualized treatment plans by to provide patients with chronic diseases. Wellframe The Future In the years and decades to come, AI will become the centrepiece of innovation in the world, ushering in growth and influencing the world in ways that pave the way for enormous possibilities as well as threats to humanity's progress, prosperity, and the very existence. Possible outcomes may include: More individuals and businesses making use of AI's capabilities and applications as it becomes more widely available and democratized. It is expected that its interaction and synergy with other technologies like cloud computing, 5G, IoT, blockchain, and quantum computing will birth new realities. Advancement to the point where it can generate its content, goods, and services without human intervention. This advancement may accelerate to the level where it will be able to learn from its own experiences and those of other AIs, leading to novel kinds of intelligence and cooperation. A greater influence, transformation, and disruption across many fields and businesses. In conclusion, the development of AI has an inevitable future. It all comes down to how researchers plan, build, implement, and regulate AI systems. How well we adjust to the new opportunities and threats that AI presents will tell how deeply AI will twist our realities. To create the kind of future we want for artificial intelligence, we must be proactive, collaborative, and inclusive.

AI Tools That Are Reshaping The Modern World Industries

Power-up: Machine Learning and Data Engineering (R)evolution for Optimizing Marketing Efforts

I want to start this guide with two important points. First, Incubators are for startups. Not ideas. Second, incubators are organisations that bet on you by investing in you in exchange for equity. If they don’t, they’re just a real estate business. I learnt these two points the hard way. Let me explain the critical mistakes I made to give you more context and ensure you avoid making them too. When I was younger, just starting my entrepreneurial journey, I took part in an incubator. But I went with an idea on a napkin. I didn’t have a fully formed vision for my product, I simply wasn’t ready. It was one of the reasons I failed. Incubators are for startups that already have a formed idea, have some market validation behind them and are ready to take it to the next step. My second mistake was going to an incubator that didn’t invest in the startups they worked with. I went somewhere that took my money in exchange for office space and free beers in the afternoon. I wasn’t working with an incubator. I was joining a co-working space. There is little value in that when you’re first building your startup. In my opinion, businesses like that shouldn’t even be called incubators. It’s vastly different from going to an actual incubator that will invest in your startup for two reasons: You’re going to learn from industry leaders who have a wealth of experience in what they’re doing and a vested interest in your success; You’re going to learn from other founders in your cohort because if the incubator thinks they’re also worth investing in they must also be doing something great. Yes, at an “incubator” where you pay for the privilege of office space you’ll also meet other founders. However, there’s no guarantee that those founders know what they’re doing because the “incubator’s” main focus is filling empty desks. Whereas an incubator's focus should be on guiding and investing in young startups that have potential. With that in mind, here’s some more information on what an incubator is, how it can help you in your entrepreneurial journey, and the list of the best ones from around the world. What is a Startup Incubator? A startup incubator is a collaborative organisation designed to support new startups that are looking to validate their formed product vision (hence the name incubator). Incubators provide a number of services designed to help you at the MVP stage – like access to startup capital. That being said, incubators aren’t for everyone. If you have access to capital and other vital resources to help you get your MVP off the ground, you may not need one – and save some equity. But we’ll delve into that later in the article. Before we get to the list of the best incubators, let’s dig a little deeper into how an incubator can help you. How a Startup Incubator Can Help You Here’s a rundown of the top three ways startup incubators help entrepreneurs succeed: 1. Mentorship There are a plethora of hurdles on the road to startup success. While some of them will be unique to your specific business, there are many that are common among all startups. If you choose to work with a startup incubator, you’ll have access to successful entrepreneurs who’ve faced many of those hurdles already. As Warren Buffet once said: It’s good to learn from your mistakes. It’s better to learn from other people’s mistakes. Connecting with these entrepreneurs will help you avoid common mistakes altogether – which will save you a lot of time as you build your MVP. They can help you with everything from business basics and presentation skills to marketing advice, product expertise and connections to strategic partners. This leads me to the second benefit, opportunities for funding. 2. Opportunities for Funding Getting your startup off the ground is an expensive endeavor. Most startup incubators will provide you with funding opportunities at key points in your collaboration with them. For example, Y Combinator offers a of $500k to every startup they work with. This investment is made on two separate safes in return for a percentage of the company. standard deal But, as I’ve already mentioned, startup incubators provide connections to strategic partners. This includes investors. Most incubators have a network of both angel investors and venture capitalists who can assist with further fundraising opportunities. 3. An Entrepreneurial Environment The final benefit of working with a startup incubator is the opportunity to work in an entrepreneurial environment. This starts with a workspace. Most incubators will offer you space in a shared co-working or even a dedicated office for you and your founding team. That being said, some of the biggest incubators don’t offer a physical space – Y Combinator is a prime example here. This agreement lasts for the duration of your collaboration with them. That being said, some incubators let you stay in the space for months or even years after the collaboration ends. Aside from the financial benefits, having a workspace surrounded by other startup founders is a great way to immerse yourself in the entrepreneurial environment. However, incubators don’t only offer this resource as part of the package. Many will also offer cloud computing credits, discounted SaaS subscriptions and other products and services to make life easier. Finally, it’s common to take part in a Demo Day as part of working with an incubator. This gives you the opportunity to showcase your product, pitch your idea and network with investors, advisors and other key stakeholders. What to Assess Before Joining an Incubator If the Model Fits Your Needs Not every startup fulfils the growth targets often demanded by startup incubators. And that’s ok, there are out there that are still very successful without seeing rapid growth. plenty of startups Instead, they focused on steady growth. Take as an example. The now giant communications company and still achieved a successful IPO. Bandwidth bootstrapped for 15 years The Amount of Equity You’re Selling Most startup incubators work off fixed equity percentages. By nature, that means they’re overpaying some companies and discounting others. Before joining an incubator, you have to see if the capital they invest is worth the percentage you’re giving up for your specific scenario. Profiles of the Mentors A large part of the benefit of joining a startup incubator is the mentors you’ll meet during the experience. If those mentors aren’t entrepreneurs themselves, with a fantastic track record avoid that incubator. In my experience, many incubators are full of wannabe angel investors who are posing as startup mentors. I can’t overstate how important it is that you don’t get trapped in such a programme. How Easy it is to Get In A quick Google will tell you how competitive the startup incubator ecosystem is. You’ll read that places are limited and there are thousands of other founders just like you fighting for a coveted spot. With that in mind, if a startup incubator is easy to get into it probably isn’t a great programme to join. Put simply, places like Y Combinator didn’t get their reputation from an easy application process. Is there “Industrial Value” in the Programme You have to evaluate how valuable the programme is to your specific industry. While seed capital is important, it might be worth going with an incubator that offers less than the average if they can introduce you to the right contacts in your industry. Let’s say you’re an aerospace startup for example. On one hand, you have an incubator that can offer a warm introduction to Nasa, Boeing, SpaceX (or SurfAir’s founder and friend of Altar, ), but only invests $40k of seed capital. Wade Eyerly On the other, you have an incubator offering $150k of seed capital, but they can’t offer you a warm introduction. The former may be more advantageous in your situation. Should You Work With a Startup Incubator? To answer this question, I recommend first looking at what you need from a startup incubator. If you think it will enhance your startup journey and the knowledge, capital, network and resources are worth giving up the equity, go for it. Just make sure you choose the right incubator for you. On the other side of the coin, if you can bootstrap your startup and already have a wealth of industry and product knowledge, good connections within your market and can find all the resources you need yourself, maybe an incubator isn’t for you. If you do go down the route of working with an incubator, make sure you get the most out of the experience possible. Take this example from a successful entrepreneur, Giacomo De Lorenzo. I sat down with him recently as part of our content series The Startup Journey: “Our incubator was extremely helpful, but not all of the startups that were part of the programme made the right use of them. They weren’t calling them twice a day. Whereas I knew I had that opportunity so I took it, and it was very valuable for us. So embrace that ability to ask for help, it’s one of the most valuable resources you have. You’ll be surprised how many people will be happy to help.” Last but not least, here’s the list of the 7 best startup incubators from across the globe. The 7 Best Startup Incubators Worldwide 1. Y Combinator Founding Date: 2005 Location: Silicon Valley, California USA Application Info: Online Application Form Funding amount: $500,000 for 7% equity Startups Funded: Over 3,000 Portfolio Companies: , , Twitch Reddit Product Hunt 2. Seedcamp Founding Date: 2007 Location: London, UK Application Info: Online Application Form Funding amount: Around £200,000 ($271,000) for 6% – 7% equity Startups Funded: 430 Portfolio Companies: , (formerly Transferwise), Revolut Wise Hopin 3. Amplify LA Founding Date: 2011 Location: Los Angeles, California, USA Application Info: Online Contact Form Funding amount: $100k for 10% equity Startups Funded: 70+ Portfolio Companies: , , (acquired by Google) FloQast Clutter Bitium 4. 500 Startups Founding Date: 2010 Location: San Francisco, California, US with locations across the globe Application Info: about locations and applications Learn more here Funding amount: $150,000 for a 6% stake Startups Funded: Over 2,500 worldwide Portfolio Companies: , , Udemy Talkdesk Canva 5. AngelPad Founding Date: 2010 Location: The US in San Francisco and New York Application Info: Apply online via landing page Funding amount: $120,000 for 7% equity Startups Funded: 150Portfolio Companies: , , Postmates Buffer AllTrails 6. Bethnal Green Ventures Founding Date: 2012 Location: London, UK Application Info: Contact the team Funding amount: £30,000 for 7% equity Startups Funded: 167 Portfolio Companies: , , Fairphone Commonplace DrDoctor 7. Techstars Founding Date: 2006 Location: Techstars is situated across 12 countries in 33 cities including London, Paris, Stockholm, Berlin, New York, LA and Toronto Application Info: You can find and apply to your location of choice . here Funding amount: $120,000 convertible note, $20,000 of which is contributed by TechStars for 6% equity Startups Funded: 2,500+ Portfolio Companies: , , Digital Ocean Chainalysis DataRobot Thanks for reading. Also Published here

How Early Stage Startups Can Leverage Startup Incubators

"War in Ukraine" is our reality and a red flag for investors. Is the war really an obstacle to the timely performance of quality work, what is the reality of the modern IT market and why is it important to work with Ukraine now? - We will tell you about this from our own experience. The history of began in 2016, when we organized the first conference on Artificial Intelligence in Kyiv. From the very idea to its implementation, it has taken us several months of hard work and some deep faith that it would be a top project. Shortly afterwards, IT recruitment, corporate data science training, data science consulting, IT outsourcing and IT outstaffing were added to our areas of work. Each direction appeared as a response to market needs. Data Science UA A few months before the full-scale invasion, we felt the tension, because Russia had driven hundreds of thousands of its troops and military equipment to our border. Like everyone else, we pushed the worst thoughts to the last. Before the Russian invasion, Ukraine became one of the world's largest centers of IT professionals working remotely for global companies. Such global IT companies as Samsung, SQUAD, DataRobot, Grammarly, Snap opened offices, developed their business in Ukraine and hired Ukrainian programmers due to high level of IT service provision, good command of the English language and  the convenient time zone. The Ukrainian IT sector can boast for impressive ratings: the first place in Eastern Europe in terms of the number of outsourced developers according to the Global Sourcing Association, the first place in Europe in terms of the number of graduates of higher technical schools, the fifth place in the world in terms of the overall quality of software developers. Since the beginning of the Russian Federation's war against Ukraine, all sectors of the Ukrainian economy have felt the negative consequences. IT was not an exception. Ukrainian IT is the third largest export industry, which makes this sector vital for economic stability. Despite the fact that IT remains almost the only industry working for export, the labor market in this area has significantly shrunk, especially for non-technical specializations. By mid-March, the market froze. Most companies have stopped hiring, but we have started actively looking for clients on the international market. Unfortunately, companies that had a sales market in Ukraine began to lay off not only technical specialists, but also developers. We threw all our forces into supporting engineers who found themselves without work and began to look for interesting projects for them on the international market, and later new companies that did not work with us before the war began to approach us. We have noticed that the number of vacancies per engineer has significantly decreased. According to statistics, the number of candidates has increased, while the number of vacancies has dropped. This statistic does not affect rare specializations that have always been relevant: Data Science, MLOPs, ML with specific skills, SW engineer (Go, Python). One of the largest IT job search resources in Ukraine — djinni — currently has 50,376 candidates. Number of vacancies: 15,160. In April 2022, DOU - the largest IT portal in Ukraine - published 2,265 offers for developers against 4,283 offers in January of the same year (a drop of 47%). The biggest drop in March-April 2022, according to DOU analytics, was recorded among vacancies for HR specialists (in March it was 76%, and in April - 71% from the number of January 2022), DBA specialists (-84% in March, −68% in April versus January), Project Managers (-69% in March and −61% in April), as well as analysts (-68% in March and −64% in April). The smallest declines were among vacancies for developers (-46% in March and -47% in April), DevOps (-50%) and design professionals (-52% in March and -48% in April). In April, the number of vacancies for C-level specialists / Architects returned to pre-war levels: 58 vacancies for them appeared, with an average of 56 vacancies per month in 2021. The median salaries of developers in vacancies have not changed much, this applies to both Senior, Middle and Junior level specialists. Median salaries for developer jobs are $2,500-$4,500 in April versus $2,500-$4,000 in January 2022. Also, there is still no difference in the salary ranges of vacancies involving work in Ukraine, abroad or remotely. Regarding the relocation of IT specialists, a large survey among the DOU community showed that 61% left their homes and moved to safer regions due to the start of the war. Almost half of IT specialists (46%) moved to a safer place in Ukraine, another 14% went abroad. Among other countries where developers left after the war, Poland is in the lead (35% of those who left abroad came here). The next most popular country is Germany (10%). Providing jobs in a "burning country" with destroyed infrastructure may seem like a risky decision, but there are good reasons to think about cooperation with Ukraine. The IT industry has grown by 36% since 2021. Uninterrupted power supply, uninterrupted Internet, and relocation of technical workers to safe regions ensure timely and high-quality provision of services even during martial law. All of our engineers are currently located in the west of Ukraine - this is a safe region of the country where there are no hostilities - and in the capital - in Kyiv. There are no problems with the network, Internet and electricity. Many embassies have already returned to Kyiv, many world-class companies (Starlink, Meta, Palantir) fearlessly enter Ukraine and plan to open their offices in Kyiv. Many companies have drawn up a Business Continuity Plan in case of unforeseen situations since the Russian invasion and annexation of Crimea in 2014. For example, using Internet services from several providers, installing backup generators in offices, providing employees with laptops with longer battery life, moving IT infrastructure to the cloud or to backups located in Poland or Latvia. During the war, our team not only closed dozens of Senior and C-level vacancies and continued to actively recruit for companies based in Europe or America, but also expanded our team of recruiters. We also have new clients, for whom we opened RnD centers in Ukraine and work on outsource and outstaff models with them. By working with Ukrainians, you bring us closer to victory. And in addition to humanitarian and military aid, cooperation with the world is important for our state. The performance of our engineers is now 100 out of 100. Some employers may have doubts about the efficiency of our engineers during the war, but from personal experience and feedback from large companies that work with Ukraine, a clear picture can be formed that Ukrainians are hardworking people who are dedicated to their work. Due to martial law and, accordingly, a reduction in the amount of entertainment and daily routine pre-war activities, employees are 100 percent engaged in work and focused on results. Engineers are additionally motivated to work at their maximum and all focus is on working and strengthening the economy, as well as providing for their families, paying taxes, financially supporting our military and bringing us closer to the inevitable victory. We hold the cyber front of our state and want to call on the world not to be afraid to cooperate with us, we will definitely not let you down. Be brave like Ukraine! While technological giants are leaving Russia and entire corporations are closing their offices and leaving the Russian market, Ukraine is developing and cementing the title of a strong and progressive state with a bright future despite enemy shelling. Cooperation with Ukraine today is a choice in the direction of a free, democratic and peaceful world. This is what we are fighting for now, protecting Europe from the imperial madness of our neighbor. We are fighting on all fronts and we call for help because we are stronger together.

Why and How to Hire Ukrainian Developers

Did you hear of the self-driving Uber car that hit and killed a woman in Arizona? On another occasion, a facial recognition solution profiled an innocent man of color as a criminal in New Jersey, and Amazon’s AI-powered recruitment tool displayed bias against female candidates. Clearly, artificial intelligence makes mistakes. Significant, even life-altering mistakes. So, how can we still get the benefits of AI while eliminating these types of errors? One option is letting human experts train, evaluate, and monitor after deployment. This concept is called human-in-the-loop (HITL) machine learning. Gartner predicts that in some industries, HITL AI solutions will . AI business solutions comprise around 30% of all automation offerings by 2025 We talked to our AI expert, , to understand how humans fit in the loop, which benefits they bring, and how to organize this process. Maksym Bochok Human in the loop definition and benefits To err is human, to really foul things up takes a computer. - Paul Ehlrich, a German physician and a Nobel Prize winner Now Ehlrich’s quote is more relevant than ever before. With AI handling critical applications, the margin for error is getting slimmer. And machines are not perfect. They build their understanding of the task based on the received training data, and can make erroneous assumptions. And this takes us to the human-in-the-loop machine learning terminology. Human in the loop means integrating human employees into the machine learning pipeline so that they can continuously train and validate models. This includes all people who work with models and their training data. How human-in-the-loop adds value to your machine learning algorithms . This is particularly important for domains that can’t tolerate errors. For example, when manufacturing critical equipment for an aircraft, we want automation and speed, but we can’t jeopardize safety. HITL is beneficial in less critical applications as well. For example, large consultancy companies that rely heavily on AI for document regulatory compliance to validate their natural language processing algorithms. Maintains a high level of precision involve human in the loop machine learning . Machine learning models during training. Moreover, they can acquire bias after deployment, as they continue to learn. Human employees can detect and eliminate this phenomenon at early stages by correcting the algorithm accordingly. Eliminates bias can become biased . ML algorithms evaluate thousands or even millions of parameters to make a final decision, and it often can’t be explained. With HITL, there is a human who understands how algorithms work and can justify the decisions they make. This is called . For instance, when a person applies for a loan and is denied, they might ask a loan officer to explain the reasoning behind the rejection and what the applicant can do to increase their chances next time. Ensures transparency explainable AI . We often hear about AI stealing people’s jobs. Machine learning with a human in the loop provides an example of how the technology can create new vacancies. Just look at the market. Opens employment opportunities Indian data annotators The role of humans in the AI pipeline Maksym explains how humans can be a part of the AI pipeline to enhance its ability to make predictions. Machine learning models operate under either supervised or unsupervised learning modes. In case of , people can perform the following tasks: supervised learning . A human employee labels the training dataset. Depending on the required expertise, this can be a domain expert or any employee with proper training. Labeling and annotation . If needed, ML engineers and programmers can make adjustments to the algorithm to make sure it can get the best out of the provided dataset. Re-engineering the model . Employees feed the model with the annotated data, view the output, make corrections, add more data if possible, and re-train the model. Training and re-training . The human in the loop machine learning lifecycle doesn’t stop after deploying the AI solution on the client’s premises. ML engineers continue to monitor its performance with the client’s consent and make adjustments to the model when required through selective verification of its output. The cases obtained through selective verification will augment the initial training dataset to improve the algorithm’s performance. Monitoring the model’s performance after deployment In , algorithms take unlabeled data as input and find structure on their own. In this case, humans do not annotate the dataset and don’t interfere much in the initial training. But they can significantly enrich the model by performing step 4 above. unsupervised machine learning When human in the loop machine learning is an absolute necessity Maksym believes that the human in the loop approach is beneficial for most machine learning use cases. AI solutions are impressive at making optimal predictions when trained on large extensive datasets, while humans can recognize patterns from a limited supply of low-quality data samples. Combining both capabilities together can create a powerful system. Even though in some applications ML models can do well with limited human intervention, there are cases where a full-blown human in the loop system is a must: When any mistake by the algorithm can be very costly, such as in . medical diagnosis When the data you need to properly train the algorithm is scarce. More training data always equates to better model performance. With the help of post-production model monitoring, you can augment the training data with relevant samples, giving the model more examples to learn from. In the case of one-shot learning when an algorithm is trained on hundreds or even thousands of samples to classify some objects. And then another class is added, and the algorithm has to learn to identify it from only a few training samples. In heavily regulated industries where it is essential to explain how the algorithms reached its conclusions. For example, when doctors use , they need to justify this treatment plan to the patient. AI to suggest personalized cancer treatments When looking at the type of data that ML algorithms process, HITL AI would be essential for and natural language processing (NLP), especially when it comes to of a text that might contain sarcasm. HITL is less important for tabular data and time series analysis. computer vision applications sentiment analysis Tips on enhancing artificial intelligence with human in the loop practices Maksym offers the following tips on how to successfully implement the human in the loop approach in machine learning: When monitoring and analyzing an algorithm’s performance after deployment, no matter how good the human in the loop system is, human participants will not be able to pay attention to every input the algorithm processes and every output it generates. Choose your cases wisely. Use selective verification to pick the cases that are worthy of your attention. Maksym suggests these approaches to smart case selection: Based on . For example, an algorithm needs to classify every input image either as a cat or a dog. The images that receive a confidence level of around 48/52 or anything similar are the ones that confuse the algorithms and need to be properly labeled and used to re-train the model. confidence levels of “trivial” cases. Let’s assume that only one out of ten cases holds valuable information when it comes to an algorithm’s performance. An example of such a case is when the model is overconfident about a wrong prediction. You should definitely consider this case, but you also need to randomly select one out of the remaining nine cases to make sure the algorithm doesn’t grow overconfident with its wrong predictions or allow bias. Random verification When analyzing the cases you picked in the previous step, don’t limit yourself to the final result. Instead of looking at the output of the final set of neurons in neural networks, check the previous layer, like in the image below, and analyze the distribution of distances between a wrong prediction and the closest correct predictions the algorithm makes. Encourage the algorithm’s end users to give feedback on its performance. Construct feedback forms and make them available to everyone, so that users can convey any concerns they may have. Keep augmenting the training dataset iteratively using data points from the previous steps. This way, you will be sure that your algorithm remains relevant even when some changes take place at the client’s operations. Off-the-shelf HITL-enabled AI tools There are some ready-made human in the loop machine learning tools that allow you to label training datasets and verify the outcome. However, you might not be able to implement the tips above with these standardized tools. Here are a few human in the loop tool examples: Google Cloud HITL This solution offers a workflow and a user interface (UI) that people can utilize to label, review, and edit the data extracted from documents. The client company can either use their own employees as labelers or can hire Google HITL workforce to accomplish the task. The tool has certain UI features to streamline labelers’ workflow and filter the output based on the confidence threshold. It also allows companies to manage their labelers' pool. Amazon Augmented AI (Amazon A2I) This human in the loop artificial intelligence tool allows people to review low-confidence and random ML predictions. Unlike Google Cloud HITL, which only operates on text, Amazon A2I can complement Amazon Recognition to extract images and validate results. It can also help review tabular data. If a client is not happy with the supplied A2I workflow, they can develop their own approach with SageMaker or a similar tool. DataRobot Humble AI Humble AI permits people to specify a set of rules that ML models have to apply while making predictions. Every rule includes a condition and a corresponding action. Currently, there are three actions: , when humans just monitor the corresponding condition without interfering No operation , when people can replace the model’s output with a different value Overriding prediction , simply discarding the prediction altogether Returning error So, is machine learning with a human in the loop the best approach for you? Employing the human-in-the-loop AI approach improves accuracy, transparency, and quality of predictions. It also increases costs and time needed to complete the task due to human intervention while creating employment opportunities, which is a positive side effect. Despite the obvious benefits of HITL AI, there are applications where human-out-of-the-loop is a preferred approach because of the risks associated with certain activities. Think of autonomous weapon development and deployment. If you feel like your ML algorithms can use a human in the loop, but you are not sure how to balance operational costs and the desired accuracy and explainability, reach out to . They will work with you to find the right fit. If human-in-the-loop machine learning is not the optimal solution in your case, there are other ML tricks that can help you overcome the problem of training data scarcity: machine learning consultants , when you fine-tune pre-trained models with your own data Transfer learning , when you use a large unlabeled dataset together with a small number of labeled samples Semi-supervised learning , when you mask a random part of the training sample in each batch and the algorithm tries to predict it Self-supervised learning Are you considering improving your ML model’s accuracy and explainability? ! ITRex AI experts will study your situation and devise an optimal human in the loop approach to address your needs. Get in touch

Using Human-in-the-Loop Approach in Machine Learning

Exploring the Top Data Science and Machine Learning (DSML) Platforms of 2022 

Entire companies are being built on top of the world's fastest-growing software. Slack, a company expecting to surpass $600 million in annual recurring revenue is still growing at an upward pace of 55% year-over-year, they have over 100,000 paying customers, and their net dollar retention is 134%. (Stats provided via ) Slack's Third Quarter Fiscal Year 2020 Results With such meteoric growth, third-party products are opting to build entire companies on top of Slack’s booming platform which recently . reached 2,000+ apps Image Via Twitter Slack is also doing its part to enable these companies to build on top of their platform. As part of their recently minted $80 million venture fund ( ), Slack has invested in the like of Onna, Troops, Lattice, Polly, and many more to enhance the Slack experience. The Slack Fund Sahil Lavingia, Founder & CEO of Gumroad and popular SaaS Twitter philosophist is bullish on Slack too. Image Via Twitter Here’s a list of 5 companies who have built on top of Slack: 1. - Giving you control over your Slack data. Onna Onna makes the process of collecting all of your companies a breeze. Native Slack capabilities don’t give you the ability to find and extract target pieces of data (known as ). Onna gives companies the ability to search through their data to find exactly what they need. Slack data Slack eDiscovery The growth of Onna has led to led by Dawn Capital and participated in by Slack themselves. fundraising Norman Fiore, General Partner at Dawn Capital, said: "Helping employees find, organize and protect crucial business information when it sits across all these apps is a big challenge. Onna's solution has already seen great traction in eDiscovery with a legal use case, with annual recurring revenue having more than tripled over the past year." "But Salim (Onna’s founder) and the team at Onna rightly recognize the opportunity is much bigger. By building an internal search engine for enterprise and giving all employees fast access to their own information, Onna can be an industry game-changer." Notable companies using Onna include: Facebook, Dropbox, Fitbit, EA, Vizio, and more. 2. Halp - Conversational ticketing for teams who love Slack. gives Slack users the power to assign, prioritize, and answer requests from Slack with ease improving response times, employee happiness, and overall productivity. Halp Co-founder and CEO Fletcher Richman created Halp when he realized that while workplaces were moving away from email-based communication and towards conversational messaging, internal ticketing remained inefficient and stuck in the past. Richman said in , “The way people work is fundamentally changing and the systems we use in the workplace need to change as well,” said Richman. Halp’s seed funding press release “Conversational messaging is the future, and with this funding, we are helping companies upgrade their legacy ticketing systems and solve internal problems faster and more efficiently. As a result, our customers have happier employees, better employee retention and more" Notable companies using Halp include: Slack, Adobe, Gusto, CarGurus, SeatGeek and more. 3. - A better way to sell for sales teams that Slack. Troops Troops makes sales a team sport - empowering people, automating workflows, and supercharging sales performance. In short, it automates routine sales management activities and creates visibility into important deals through integrations with employee emails and Salesforce. Troops founder and chief executive officer Dan Reich, who previously co-founded TULA Skincare, told he opted to build a Slackbot rather than create an independent platform because Slack is a rocket ship and he wanted a seat on board: TechCrunch “When you think about where Slack will go in the future, it’s obvious to us that companies all over the world will be using it,” he said. Troops has raised $12 million in Series B funding in a round led by Aspect Ventures, with participation from the Slack Fund Notable companies using Troops include: Slack, HubSpot, Intercom, WeWork, Looker, Lyft, Square, InVision, and more. 4. - The best way to measure your most critical workflows in Slack. Polly Whether you’re on Slack or Microsoft Teams, Polly helps you get timely feedback from the right people, at the right time, in the right place. “The Polly team has taken a simple idea – surveys inside of Slack – and created a whole new way of collecting people-powered data on modern communication platforms,” commented Sudip Chakrabarti, Partner, Madrona Venture Group in Polly’s . Series A funding announcement “As modern workflows move to real-time messaging platforms, Polly enables enterprises to get a holistic picture of their customers, employees, product, and brand.” Polly raised $7 million in Series A funding led by Madrona Venture Group with participation from the Slack Fund. Notable companies using Polly include: Slack, Auth0, PureStorage, DataRobot, and more. 5. - Develop engaged and high-performing teams. Lattice Bring more actionable feedback, praise, and awareness to everyday Slack conversations with Lattice. “Performance management can feel like extra work. Frequently, employees need multiple reminders to fill out a performance review, forget to give each other feedback, and never set goals. Employees want to use HR software that feels and works like the apps they use in the rest of their lives — this dynamic exists in every organization.” Lattice said in the blog post discussing their . Slack integration If your team is like Slack’s, or any of their millions of users, you are living in the Slack app every day. Now, you won’t ever need to leave , turn on a constant stream of recognition or give helpful nudges to your team when they need to do something for you (like writing that peer review). Slack to give your teammates feedback and praise $25 million in Series C funding led by Tiger Global Management with participation from the Slack Fund. Lattice raised Notable include: Slack, Reddit, Asana, Cruise, Samsara, Button, Clio, and more. companies using Lattice NOTE It makes sense to want to customize your Slack as much as possible, but for free accounts, you can only add up to 10 Slack apps. If you want more, you’ll need to upgrade your Slack plan. To get more insights on Platforms, APIs, Web3, and more — subscribe to . The API Economy of SaaS RESOURCES Why we’ll see more startups built on platforms like Slack Building a business on Slack Best Slack apps and bots to add to your workspace (not Slack related, but an incredibly powerful essay on why platforms like Shopify and Slack are so popular Shopify and the Power of Platforms

The Best Companies Built On Top Of Slack

Big data is a well-known term in the business industry for more than a decade now and big data analytics have also been around for quite some time. All businesses want to utilize big data in a positive way and perform big data analytics to help data analysts get the most out of the data wealth of the company. Management can make informed decisions with more confidence on the basis of big data analytics. Big data analytics can be applied for all and any business to boost their revenue and conversions and identify common mistakes or faulty practices that might be affecting their growth in a negative way as compared to their competitors. The data insights from big data analytics can help businesses overcome this gap and make structural and strategic changes to their business operations and procedures for a positive impact and business growth. “Without big data analytics, companies are blind and deaf, wandering out onto the web like deer on a freeway.” – , Author and Consultant. Geoffrey Moore What is Big Data? Big data is a term used to refer to and is streaming in in a business. These data may be coming from different data sources and are managed by different database systems set up across the business organization. Data is often collated and stored in a compatible format before big data analytics could be performed on the business data. business data that is spread across the organization from multiple channels Big data analytics help businesses aggregate, manage, and process their data and use it to answer important questions and identify opportunities. It can help them to manage their operations more efficiently and lead to more productivity, higher performance, greater profits, and better customer management and relations. Application Industries for Big Data Analytics Government: Big data applications are being devised for government agencies to manage their large databases that hold sensitive data for different types of government services like identification documents, citizens’ data, and data from different government agencies. All this data is interlinked and comes from completely different sections or offices of government organizations. Healthcare: Healthcare is another area of application that encompasses big bulks of data and information coming from different sources. Areas like patients and hospital data, medicine and pharmaceutical data, medical research and scientific data, and health insurance data all come under this umbrella. Education: Education has expanded a lot over time and falls in the application of big data. Information about students, parents, teachers, educationalists can be a focus of education-related big data. Also included in this area are the SEN, EAL, and gifted and talented children data that is dealt with separately in a school environment. Media: Media and entertainment are a big industry that is a main application of big data analytics. Scores of social media platforms and music and video websites are available that are generating a huge amount of data that is completely out of the scope of traditional database management systems and data reporting and analysis tools. Banking: Banking is another enormous industry that utilizes big data analytics to manage its large wealth of big data generated by different banking sectors and expert areas. The online banking systems that offer online services contribute their share to the big data in the banking industry. New banking services and products are coming into the market every day and they bring with them their own fair share of big data and the volume keeps increasing. This big data is managed and studied using big data analytic tools and techniques. Importance of Big Data Analytics Gaining insight into the types of industries involved in the big data scenario and the bulk of data that is being produced by these industries give us an idea about how important it is to convert this bulk of big data to meaningful insights and actionable visualizations that can provide data analysts means to understand and coerce the data trends and use them to impact the business positively. Some advantages of big data analytics are cost control and able to increase productivity and performance by analyzing the business data and reaching conclusions about key performance metrics. Big data analytics also allows data analysts to convince management into making informed decisions quicker and with more confidence. It also has a positive impact on customer relationships and helps to retain customers by acting and improving services based on customer feedback. Tools and Techniques for Big Data Analytics In this section, we will discuss some tools and techniques that form the basis of big data analytics and helps data analysts perform their tasks quickly and with ease. These big data analytics tools can help aggregate, assemble, and organize data before applying algorithms to understand the data trends. Machine learning Machine learning is a set of algorithms designed to run on big data and identify the data patterns for performing analytics. Machine learning is a subset of data science, which lets data analysts study big data and forecast data metric that helps the management take short-term and long-term decisions regarding their operations and services. Predictive analysis Predictive analysis of big data allows business analysts to perform detailed data analytics for the big data of the business to predict future outcomes based on historical data trends. These predictions are very useful for customer engagements and suggesting customized adverts according to their interests. A range of predictions can be made for different business products and services and action plans can be developed to improve services that may in turn increase revenue. Data mining Data mining tools implement algorithms to identify data patterns and create visualizations that help data analysts to answer business-related questions and make predictions about business performance. Data mining can be useful when a business is dealing with bulks of data that need refinement and organization for detailed study. Data Querying Data querying is another tool to apply big data analytics for converting data into information. Data analysts perform analytics to answer queries that target increasing performance metrics and specific questions related to business data. Data querying is important when management wants answers to specific questions and data analysts study data to specifically answer these performance-related questions. Data visualization The importance of data visualization can never be negated in big data analytics as visuals are a powerful tool to understand data and derive information and statistics that can lead to changes in course of action and business procedures to increase productivity and improve services. Data reporting and analytics softwares can achieve this goal by offering features to create customized visualizations and data reports. help business analysts Popular Software Tools for Big Data Analytics Lots of big data analytics tools and softwares are making their way into the software market every day. They are packed with features and options to perform big data analytics. BigML and DataRobot are more inclined towards using artificial intelligence and machine learning techniques for big data analytics. Many business intelligence softwares and data reporting tools like Tableau, Sisense, and provide enterprise-level businesses with features to perform big data analytics through data visualization, data reporting, and analytics. dotnet report builder Before a business makes a decision to purchase a software solution for big data analytics, the data analysts’ must perform a requirement analysis for their business to identify the data analytics needs for their business. Then, they can search the market for big data analytic solution that best matches their needs and blends well with their existing business workflow.

What Is Big Data? Understanding The Business Use of Big Data Analytics

AI and there’s lots of money to be made from it. But organisations keep making the news over AI governance failings, such as Microsoft’s and google images . We’re seeing a growth of ethics and governance councils but with mixed success - Google . is the future chatbot that turned racist labelling African-Americans as gorillas shut theirs down Why is good governance proving so difficult? Does there have to be a trade-off between good governance and innovation? The quick answer on this is that For the detailed answer, read on. running AI with governance poses tough technical challenges. The solutions fall under the emerging field of MLOps... but it is an emerging field. The AI Rush AI is taking off. The market for AI is from $9.5bn in 2018 to $118.6bn by 2025. Naturally there is a race to get to the opportunities first. has been forecast to grow However, making money from machine learning right now is not easy. Up to of machine learning projects never go live. Some projects fail because the idea doesn’t work or the data isn’t available. Even after those hurdles, the journey to running the model live isn’t easy. 87% There can also be demanding hardware needs. The data flowing through the model and the quality of predictions can require monitoring. All this makes for a . Getting data, preparing it, keeping it fresh and managing changes to it are all challenging. challenging AI operations landscape One approach is the way - just deal with the issues as they come. This is especially risky with machine learning as things can break in a whole range of ways. Let’s look at some of them. “move fast and break things” Challenges for Running AI in Prod Machine learning models are only as good as the data that they are trained on. Machine learning takes patterns from known data and reapplies the extracted patterns to new data. If the known data is not representative of the new data then the model will give bad predictions. So a major challenge is getting data that is representative. The level of quality required of predictions varies with use-case. If the predictions are going to be followed up by a human (e.g. flagging transactions as potential fraud risks) then a poor prediction might not be too serious. If a self-driving car then that could be very serious. In some cases even one bad prediction could be intolerable. mistakes a stop sign for a speed limit sign Outliers Sometimes a model can perform well but occasional predictions might go astray. This is likely to happen if there are particular data points which stand out from others - : outlier data points Outliers don’t fit the pattern of the rest of the data and the model is built to be general so it will predict on them as though they were within the main pattern. If you’ve got outliers that you can detect then you may be able to plan for that and send them down a different process. Naturally it takes time to get a process in place so it’s a risk you want to mitigate in advance rather than try to react to. Concept Drift You may get good representative data that you train your model on and it might perform well in live. Then it might mysteriously start to perform worse and worse. This can happen if the relationships for the data change in the real world. For example, if you’re predicting what fashion items are likely to sell and the season changes. Then you’ll find your model suggesting swimwear in winter. To avoid losing money you’d need to scramble to get fresh data to train your model on. Bias There are data points that might correlate highly with certain outcomes but which we shouldn’t use for ethical reasons. For example, we . wouldn’t find it acceptable to use race in automated recommendations for parole An apparent model bias situation hit the news at the end of 2019 with , despite his wife having a higher credit score. Apple co-founder Steve Wozniak was among those to report the same: AppleCard offering David Heinemeier Hanson a much higher credit limit than his wife The operators of AppleCard have stated that they didn't use gender as a data-point. However, gender could enter indirectly. For example, if occupation is used this could . indirectly bias for gender in some cases as certain occupations are dominated by one gender (e.g. primary school teachers tend to be female) The AppleCard case has attracted regulatory attention and highlights that . New York's Department of Financial Services an “algorithm that intentionally or not results in discriminatory treatment of women or any other protected class violates New York law.” bias represents legal risk as well as a reputation risk stated that Privacy The has brought a lot of attention on privacy issues. A key part of that was the sharing of data without adequate consent from the person that the data concerned. There can be risks in this space for machine learning. Facebook and Cambridge Analytica scandal A machine learning model is likely to predict similarly to the nearest data points in the training data. Say you’re predicting voting and somebody asks for predictions for retired female voters in a given district. One might not expect that to reveal much about who was surveyed for the training data - but . It might then be possible for somebody so-inclined to work backwards and figure out things about your data that you didn’t know you were revealing. it might if there’s only a handful of retired female voters in that district Range of Risks So we’re seeing a range of risks. Poor data might lead to poor predictions that cost money. Or poor predictions might be outright dangerous. Both happened with the which had $62M of funding and was shut down after making too many unsafe predictions. failed Watson for Oncology project Even if the training data is representative for a wide range of cases, the real world might lead to unanticipated kinds of data. This happened with and the . apple’s face recognition software that was tricked by a mask self-driving car that wasn’t able to recognise jaywalkers Machine learning models can be briefly successful and then get tripped up when the real-world data changes (concept drift). But models can also be too responsive to new data. was learning continuously from its interactions online and the result was that it became racist. Microsoft’s famous chatbot It’s also possible to get into hot water by overlooking the ethical dimensions of using particular data points such as gender or race. An example of this was Amazon Rekognition far more likely to mistake a member of US Congress for a criminal (compared with criminal mugshots) if they were a person of colour: being shown by ACLU to be With increasingly high-profile incidents hitting the news, organisations are concerned to ensure that AI adoption doesn’t come at an unacceptable cost. Striving for Good Governance Regulation Government and regulators are also concerned. Facebook has even gone so far as to . We are in the early days of as regulators have concerns about what , especially on innovation. But there are some notable attempts emerging. call on government to provide more regulation forming regulation effects greater regulation might have The European Union’s “the data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her.” It adds that data subjects have a right to “meaningful information about the logic involved” and to “the significance and the envisaged consequences” of automated decision-making. GDPR legislation states that The general thrust is that if AI is used to make a decision that impacts on me then I should be able to challenge that decision and get a meaningful explanation. If concern is raised about the process then the organisation might be challenged on whether the process introduced unnecessary risk. Guidelines have been issued by the US Federal Reserve for the financial industry on the use of quantitative models ( - ). These also expect organisations not to take unnecessary risk and to be able to justify their design decisions. Furthermore, they call for active monitoring and evaluation of models. The management of risk involved in the design and use of a model should be able to stand up to challenge by an informed party and the institution is expected to have processes in place to ensure that the risk trade-offs of any design are justifiable. SR11-7 pdf Internal AI Ethics Bodies Part of the AI governance picture now is organisations setting up councils to oversee guidelines on appropriate use of AI in their organisation. A key problem these councils face is how to between high-level aspirations like transparency and fairness and actionable advice. close the gap Closing the Gap So how to close the gap between governance aspirations and data science practice? And can it be done in a way that doesn’t compromise on innovation? Lessons of DevOps It is possible to go faster and with better governance. This is one of the lessons of the . Automating processes allows them to be executed faster and more predictably and for more reliable tracking. There may be an upfront cost to automating deployment and monitoring processes but the cost is paid back by the benefits of being able to make iterative improvements more often, . rise of DevOps gathering more feedback and responding to feedback more quickly DevOps for Machine Learning is Special Unfortunately are needed in order to close the gap between AI governance aspirations and data science practice. more than just existing DevOps practices Reproducibility If something goes wrong with a model then you might naturally want to be able to go back to how it was trained (that particular version of it) and make a tweak and train it again. If your process is to stand up to challenge by a potential regulator/auditor then you’ll need to think about whether they’ll ask for for this. The practice of being able to go back to a particular version and rebuild it from source is pretty typical for mainstream DevOps but it’s not an easy thing to achieve for machine learning. There can be a lot of data involved, it can change often and it can be transformed as it goes through the process. An : example machine learning pipeline might have several stages before generating a model For reproducibility you would want to be able to go back to a particular training run that resulted in a particular model (the .pkl here) and find all the versions of the source files (the .py files) and the versions of the data, as well as who/what initiated the run and when and any metrics (e.g. accuracy). Monitoring For full reproducibility you would also need to know exactly what the request was that was made and what the prediction was. So this data needs to be stored and made retrievable. Full logging of predictions is useful because it gives you: An audit trail of requests that can be individually investigated. Fresh data that can be used for training the next iteration of the model. A stream of data that can be monitored against benchmarks. Monitoring of the data stream could be used to mitigate the risk of concept drift. This would involve watching whether data is close enough to the training data for predictions to be trustworthy. Alerts could be set up for if the data strays too far. Deployments Deploying a new iteration of a model to take over all of live traffic in one go might be risky unless you’re entirely sure that live traffic matches the training data. Given the risks, you might want to tentatively deploy new versions of models to just run against a portion of live data before scaling up to take over from the old version. Explainability We’ve seen that GDPR asks that data subjects have a right to “meaningful information about the logic involved.” Providing explanations for machine learning decisions can be challenging. Machine learning takes patterns from data and reapplies them but the patterns are not always simple lines that can be visualized. There are a range of types of technique and the . kind of explanations that are achievable can vary There’s an example of the use of one such technique . In that example income is predicted from US census data which includes features such as age, marital status, gender and occupation. A technique called ‘anchors’ is used to reveal which feature values were most relevant to a particular prediction. The technique scans the dataset and finds patterns in the predictions. A particular low-income prediction is chosen and the technique is used to reveal that a separated female would be classified as low-income in 95% of cases, meaning that other features (such as occupation or age) then have very little relevance for separated females. in Seldon’s Alibi explainer library The income classifier example bears a striking resemblance to the incident with . Building explainability into the design would’ve allowed it to be determined quickly whether gender was the reason for the bias. AppleCard offering David Heinemeier Hanson a much higher credit limit than his wife There are also explanation techniques applicable to text classification or image predictions. Another shows how particular sections from an image can be determined to be especially relevant to how it was classified: Alibi example Here the nose-region of the picture was especially important for determining that the ImageNet picture was of a Persian Cat. Rise of MLOps We can now see that there’s a range of challenges to tackle for good AI governance. Many organisations are choosing whether to put together an MLOps infrastructure in-house to tackle these challenges or to choose a platform. ML Platforms ML Platforms come in a range of flavours. Some are part of a cloud provider offering, such as AWS SageMaker or AzureML. Others are an offering in themselves, such as Databricks Mlflow. The level of automation involved in platforms can vary. For example Google’s AutoML and DataRobot aim to enable models to be produced with minimal machine learning expertise. Some platforms are closed source, some are open (such as the kubernetes-oriented kubeflow). Some are more oriented towards particular stages of machine learning (e.g. training vs deployment) and the extent of support can vary when it comes to aiding reproducibility, monitoring or explainability. Organisations have to consider their use-cases and priorities when evaluating platforms. Tool Landscape Rather than choosing an existing platform, organisations or teams can choose to assemble their own. There is not yet a ‘ ’ of obvious choices. Instead there’s a range of choices for each part of the pipeline and a particular assemblage of choices might not have been designed to work together. Teams have to evaluate tools for how well they fit their needs and also how well they fit their other tool choices. canonical stack The represents an interesting position here as it aims to bring together best-of-breed open source tools and ensure they work well together. , where I am an Engineer, specialises in open source tools for and . We partner with kubeflow on tools such as . We also offer , an enterprise product that brings open source tools together to get the best out of them and speed up project delivery. Seldon Deploy can be used either standalone or together with other platforms to add an extra layer of visibility and governance. kubeflow platform Seldon deployment governance KFServing Seldon Deploy Industry Initiatives The aims to foster open source collaboration on machine learning tools. It has a great . The full landscape is large but here’s a snippet: Linux Foundation’s LF AI project visualization of the tool landscape The is a volunteer-led research centre with and has put a particular focus on providing practical tools and showing how to follow its . The Institute has created an , a framework to and listings of . Institute for Ethical AI over 150 expert members principles AI explainability library evaluate machine learning suppliers MLOps tools The was recently formed by Dan Jeffries of (a company specialising in data pipelines, versioning and lineage). The foundation has put an emphasis on closing the gap between AI governance aspirations and realisation. They’ve emphasised auditing and explainability as well as the formation of AI QA/response teams. Practical AI Ethics Foundation Pachyderm suggesting This is of what’s going on in this fast-moving space. just some Move Fast with Stable Infrastructure Part of the future for machine learning is to follow . MLOps infrastructure will enable organisations to go faster and with greater governance. Right now MLOps is a space of innovation. Projects looking to leverage MLOps have to first learn and take advantage of it in a way that best suits their situation and aims. Facebook’s “move fast with stable infrastructure” motto how to navigate the space (Title image from , attributed to Facebook.) https://medium.com/swlh/move-fast-and-break-things-is-not-dead-8260b0718d90

Stories, News & Mentions about DATAROBOT

Datarobot’s Stories on HackerNoon

Datarobot's News & Mentions