Keeping up with an industry as fast-moving as\u00a0AI<\/a>\u00a0is a tall order. So until an AI can do it for you, here\u2019s a handy roundup of recent stories in the world of machine learning, along with notable research and experiments we didn\u2019t cover on their own.<\/p>\n By the way, TechCrunch plans to launch an AI newsletter soon. Stay tuned. In the meantime, we\u2019re upping the cadence of our semiregular AI column, which was previously twice a month (or so), to weekly \u2014 so be on the lookout for more editions.<\/p>\n This week in AI, OpenAI once again dominated the news cycle (despite Google\u2019s best efforts) with a product launch, but also, with some palace intrigue. The company unveiled GPT-4o, its most capable generative model yet, and just days later effectively disbanded a team working on the problem of developing controls to prevent \u201csuperintelligent\u201d AI systems from going rogue.<\/p>\n The dismantling of the team generated a lot of headlines, predictably. Reporting \u2014 including ours<\/a> \u2014 suggests that OpenAI deprioritized the team\u2019s safety research in favor of launching new products like the aforementioned GPT-4o, ultimately leading to the resignation<\/a> of the team\u2019s two co-leads, Jan Leike and OpenAI co-founder Ilya Sutskever. <\/p>\n Superintelligent AI is more theoretical than real at this point; it\u2019s not clear when \u2014 or whether \u2014 the tech industry will achieve the breakthroughs necessary in order to create AI capable of accomplishing any task a human can. But the coverage from this week would seem to confirm one thing: that OpenAI\u2019s leadership \u2014 in particular CEO Sam Altman \u2014 has increasingly chosen to prioritize products over safeguards. <\/p>\n Altman reportedly \u201cinfuriated<\/a>\u201d Sutskever by rushing the launch of AI-powered features at OpenAI\u2019s first dev conference last November. And he\u2019s said to have been<\/a>\u00a0critical of Helen Toner, director at Georgetown\u2019s Center for Security and Emerging Technologies and a former member of OpenAI\u2019s board, over a paper she co-authored that cast OpenAI\u2019s approach to safety in a critical light \u2014 to the point where he attempted to push her off the board. <\/p>\n Over the past year or so, OpenAI\u2019s let its chatbot store fill up with spam<\/a> and (allegedly) scraped data from YouTube<\/a> against the platform\u2019s terms of service while voicing ambitions to let its AI generate depictions of porn<\/a> and gore. Certainly, safety seems to have taken a back seat at the company \u2014 and a growing number of OpenAI safety researchers have come to the conclusion that their work would be better supported elsewhere. <\/p>\n Here are some other AI stories of note from the past few days:<\/p>\n AI safety is obviously top of mind this week with the OpenAI departures, but Google Deepmind is plowing onwards with a new \u201cFrontier Safety Framework.\u201d<\/a> Basically it\u2019s the organization\u2019s strategy for identifying and hopefully preventing any runaway capabilities \u2014 it doesn\u2019t have to be AGI, it could be a malware generator gone mad or the like.<\/p>\n The framework has three steps: 1. Identify potentially harmful capabilities in a model by simulating its paths of development. 2. Evaluate models regularly to detect when they have reached known \u201ccritical capability levels.\u201d 3. Apply a mitigation plan to prevent exfiltration (by another or itself) or problematic deployment. There\u2019s more detail here<\/a>. It may sound kind of like an obvious series of actions, but it\u2019s important to formalize them or everyone is just kind of winging it. That\u2019s how you get the bad AI.<\/p>\n A rather different risk has been identified by Cambridge researchers, who are rightly concerned at the proliferation of chatbots that one trains on a dead person\u2019s data in order to provide a superficial simulacrum of that person. You may (as I do) find the whole concept somewhat abhorrent, but it could be used in grief management and other scenarios if we are careful. The problem is we are not being careful.<\/p>\n \u201cThis area of AI is an ethical minefield,\u201d said lead researcher Katarzyna Nowaczyk-Basi\u0144ska<\/a>. \u201cWe need to start thinking now about how we mitigate the social and psychological risks of digital immortality, because the technology is already here.\u201d The team identifies numerous scams, potential bad and good outcomes, and discusses the concept generally (including fake services) in a paper published in Philosophy & Technology<\/a>. Black Mirror predicts the future once again!<\/p>\n In less creepy applications of AI, physicists at MIT <\/a>are looking at a useful (to them) tool for predicting a physical system\u2019s phase or state, normally a statistical task that can grow onerous with more complex systems. But training up a machine learning model on the right data and grounding it with some known material characteristics of a system and you have yourself a considerably more efficient way to go about it. Just another example of how ML is finding niches even in advanced science.<\/p>\n Over at CU Boulder, they\u2019re talking about how AI can be used in disaster management. The tech may be useful for quick prediction of where resources will be needed, mapping damage, even helping train responders, but people are (understandably) hesitant to apply it in life-and-death scenarios.<\/p>\n Professor Amir Behzadan <\/a>is trying to move the ball forward on that, saying \u201cHuman-centered AI leads to more effective disaster response and recovery practices by promoting collaboration, understanding and inclusivity among team members, survivors and stakeholders.\u201d They\u2019re still at the workshop phase, but it\u2019s important to think deeply about this stuff before trying to, say, automate aid distribution after a hurricane.<\/p>\n\n
More machine learnings<\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/05\/manana.jpg?w=680\")
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/05\/cu-boulder-ai-workshop.webp?w=680\")
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/05\/CADS-Image.png?w=680\")