Discover more from Last Week in AI
Last Week in AI #213: US police used Clearview AI 1 million times, Bing Chatbot gets ads, Midjourney ends free trials due to abuse, and more!
Clearview database has over 30 billion images, used 1 million times by police, Bing Chat is incorporating more ads, Midjourney ends free tier to curb deepfakes
In a recent interview, Clearview AI CEO, Hoan Ton-That, told BBC that the facial recognition company now has 30 billion images scraped from online platforms like Facebook and that it estimates the service has been used almost a million times by law enforcement (the million number has not been independently verified). Clearview has been fined in Europe and banned from selling its services to US companies, but law enforcement is exempt from this ban. In an interview with Miami Police, the department says that they use Clearview for all types of crime, from murders to shoplifting. Critics are concerned about both how Clearview amassed a large database without consent from individuals and that its product has led to wrongful arrests, something the company attributes to “poor policing.” The CEO does not want to testify in court about the product’s accuracy.
Our take: We’ve been highlighting Clearview AI’s stories for years now, and it still seems incredible that a private company was able to scrape billions of images of people and sell this to law enforcement without essentially any real pushback. According to this interview, the company’s doing quite well, despite numerous high-profile cases of wrongful arrests. This does make it seem like once a technology is widely adopted, it’ll be really hard to “rein it in,” despite clear ethical concerns. We should keep in mind Clearview’s success story when considering other current and upcoming AI products that have similar ethical qualms.
Users have been spotting inline ads in responses from Microsoft Bing’s chatbot. The company is still testing various strategies and formats for these ads, and for now, the experience remains highly variable across users. Currently, we don’t have a good idea of the volume of ads served on Bing, but it suffices to say the company is eager to explore monetization strategies as it rapidly gains new users for Bing.
Our take: It’s unlikely that companies like Microsoft or Google will publish the unit economics on serving search queries via LLM-powered chat as well as the effectiveness of serving ads to profit off of chatbot traffic. Nonetheless, the fact that Microsoft is exploring this strategy means that there is a chance that this will eventually make money, even if LLMs are expensive right now. What’s neat about monetization via chats is that chat provides a much more streamlined user interface not just for information communication but as well as for actions - chatbots can easily integrate personalized and customizable buttons for commercial purposes (e.g. subscriptions, purchases, etc), and we can imagine one-day companies integrate such buy actions directly into LLM-powered chat, which will provide far greater commercial value than serving ads.
Midjourney is ending its free trial after high-profile deepfakes, created using the tool, went viral last week. These include images of Donald Trump getting arrested and Pope Francis wearing coats. The company has come under scrutiny as there is a real concern that its service could be used to spread misinformation. Other image-generating AI products tend to have stricter content restrictions. For example, OpenAI bans images of ongoing political events, conspiracy theories, politicians, hate, sexuality, and violence. This problem is exacerbated now because Midjourney’s new image-generating AI produces images that are much more realistic than before, making it difficult to immediately spot AI generations but much easier to generate misinformation.
Our take: The days of being able to easily tell if an image is AI-generated or not may be over with Midjourney v5. Misinformation from generative AI is a real concern. While some may make the “photoshop” argument - we’ve had photoshop that could doctor images for a long time, and it really hasn’t been a huge issue - generative AI is fundamentally different because of how much lower the barrier to entry is to produce vast amounts of realistic and specific fake images. In addition, there may not be any real technical solutions here, whether for prevention or detection. I think society needs to adapt to a reality where by default the authenticity of media is not to be trusted and adopt decision-making practices that are robust to these deepfakes.
Cerebras Publishes 7 Trained Generative AI Models To Open Source - "The AI company is the first to use Non-GPU tech to train GPT-based Large Language Models and make available to the AI community. The early days of an open AI community, sharing work and building on each other’s success, is over."
Researchers at the Cognition and Language Development Lab tested three- and five-year-olds to see whether robots could be better teachers than people - "Who do children prefer to learn from? Previous research has shown that even infants can identify the best informant. But would preschoolers prefer learning from a competent robot over an incompetent human?"
Professor fills gaps found in biomedical machine learning - "The presence of machine learning has been growing steadily, making itself known in our everyday technology. Some may call it “too invasive,” but machine learning can assist with learning new ways to help those who are ill. Dr."
A Wharton professor gave A.I. tools 30 minutes to work on a business project. The results were ‘superhuman’ - "Ethan Mollick, a management professor at the Wharton School of the University of Pennsylvania, has been closely following developments in generative A.I. tools, which can create essays, images, voices, code, and much else based on a user’s text prompts. "
ChatGPT and AI might have a future as your portfolio manager, study suggests - "The proliferation of artificial intelligence programs like ChatGPT and Alphabet's BardAI has already made big waves in financial markets, and a new study suggests that one day those programs may be able to trade in those markets all on their own."
Chinese creators use Midjourney’s AI to generate retro urban “photography” - "China Report is MIT Technology Review’s newsletter about technology developments in China. Sign up to receive it in your inbox every Tuesday. Welcome back to China Report!"
Machine learning to the rescue: Preventing cyberbullying in real time - "In today's digital age, the widespread use of social media and online communication has brought new challenges, including the rise of cyberbullying.
"New in-home AI tool monitors the health of elderly residents - "Engineers are harnessing artificial intelligence (AI) and wireless technology to unobtrusively monitor elderly people in their living spaces and provide early detection of emerging health problems."
Google Partners with AI Startup Replit to Take on Microsoft’s GitHub - "Alphabet Inc.’s Google is striking a partnership to combine its artificial intelligence language models with software from startup Replit Inc. that helps computer programmers write code, a bid to compete with a similar product from Microsoft Corp.’s GitHub and OpenAI."
Generative AI set to affect 300 million jobs across major economies - "The latest breakthroughs in artificial intelligence could lead to the automation of a quarter of the work done in the US and eurozone, according to research by Goldman Sachs."
Apple acquired a startup using AI to compress videos - "Apple has quietly acquired a Mountain View-based startup, WaveOne, that was developing AI algorithms for compressing video. Apple wouldn’t confirm the sale when asked for comment."
AI Search Startup Raises $26 Million to Offer Rival to Google - "Perplexity AI is part of a wave of startups trying to remake search with artificial intelligence. Perplexity is part of a growing wave of startups seeking to use artificial intelligence to loosen Google’s hold on online search."
Google's Shuffles Assistant Leadership to Focus on Bard AI - "Google is saying goodbye to a longtime Assistant veteran and putting its Assistant engineering vice president on Bard AI chatbot duties in a move that seems to indicate that priorities at the company are now shuffling to AI. The news comes via a leaked internal memo obtained by CNBC."
Founders of AI company NtechLab say they resigned over projects in Russia - “The founders of artificial intelligence company NtechLab said they resigned over disagreements with the company’s management and investors about projects in Russia."
Yeah, of course, YC’s winter class is oozing with AI companies - "Being an artificial intelligence company has become the soup du jour of startup land. Companies are scrambling to either incorporate AI into their existing business model or change up their marketing so whatever they were already quietly using AI to do is front and center."
The ChatGPT King Isn’t Worried, but He Knows You Might Be - "I first met Sam Altman in the summer of 2019, days after Microsoft agreed to invest $1 billion in his three-year-old start-up, OpenAI. At his suggestion, we had dinner at a small, decidedly modern restaurant not far from his home in San Francisco."
Publishers Prepare for Showdown With Microsoft, Google Over AI Tools - "Since the arrival of chatbots that can carry on conversations, make up sonnets and ace the LSAT, many people have been in awe at the artificial-intelligence technology’s capabilities. Publishers of online content share in that sense of wonder."
OpenAI CEO Sam Altman calls Elon Musk a ‘jerk’ as report says the Tesla CEO was ‘furious’ about ChatGPT’s success - "A month later, Musk again tweeted about OpenAI’s finances: “I’m still confused as to how a non-profit to which I donated ~$100M somehow became a $30B market cap for-profit."
The AI industry really should slow down a little - "What a difference four months can make. If you had asked in November how I thought AI systems were progressing, I might have shrugged. Sure, by then OpenAI had released DALL-E, and I found myself enthralled with the creative possibilities it presented."
Can a Machine Know That We Know What It Knows? - "Mind reading is common among us humans. Not in the ways that psychics claim to do it, by gaining access to the warm streams of consciousness that fill every individual’s experience, or in the ways that mentalists claim to do it, by pulling a thought out of your head at will."
How to Become an Expert on A.I. - "Welcome to On Tech: A.I., a pop-up newsletter that will teach you about artificial intelligence, especially the new breed of chatbots like ChatGPT — all in only five days. We’ll tackle some of the big themes and questions around A.I."
I Robot: the rise of AI, machine learning and ChatGPT in legal - "Paul Longhurst writes… Following on from article 17 in this 3Kites' Navigating Legaltech series, I wanted to look at the impact of AI on legal work and the potential challenges that it introduces for law firms."
AI Spring? Four Takeaways from Major Releases in Foundation Models - "As companies release new, more capable models, questions around deployment and transparency arise."
WGA Would Allow Artificial Intelligence in Scriptwriting, as Long as Writers Maintain Credit - "The Writers Guild of America has proposed allowing artificial intelligence to write scripts, as long as it does not affect writers’ credits or residuals."
FTC Is Reviewing Competition in Artificial Intelligence - "The US Federal Trade Commission is paying close attention to developments in artificial intelligence to ensure the field isn’t dominated by the major tech platforms, Chair Lina Khan said Monday."
UK to avoid fixed rules for AI – in favor of ‘context-specific guidance’ - "The UK isn’t going to be setting hard rules for AI any time soon. Today, the Department for Science, Innovation and Technology (DSIT) published a white paper setting out the government’s preference for a light-touch approach to regulating artificial intelligence."
UN agency calls on governments to implement global ethical framework for AI - "The United Nations Educational, Scientific and Cultural Organization (UNESCO) called Thursday for countries to implement its global ethical framework immediately following pleas by more than a thousand tech workers for a pause in the training of the most powerful artificial intelligence (AI) systems"
Law, Policy, & AI Update: Does Section 230 Cover Generative AI? - "This month's legal and policy developments have brought both challenges and opportunities for the AI industry. GPT-4 clearly has new capabilities––including allegedly passing the bar exam––and AI is rapidly being deployed throughout legal contexts."
Google C.E.O. Sundar Pichai on the A.I. Moment: ‘You Will See Us Be Bold’ - "In an extended interview, Mr. Pichai expressed both optimism and worry about the state of the A.I. race."
AI-generated images of Pope Francis in puffer jacket fool the internet - "People online praised Pope Francis' style this weekend after images of him wearing trendy outerwear went viral. But the pictures of him in a white puffer jacket are actually fake."
Copyright © 2023 Skynet Today, All rights reserved.