Keeping up with the rapidly evolving AI industry is no small feat. Until AI can handle that for you, here’s a quick roundup of recent stories in the world of machine learning, along with notable research and experiments that we didn’t cover separately.
By the way, TechCrunch will soon launch an AI newsletter. In the meantime, we’re increasing the frequency of our AI column from twice a month to weekly, so expect more frequent updates.
OpenAI's Dominance and Shifting Priorities
This week, OpenAI once again captured headlines with a new product launch and some internal changes. The company unveiled GPT-4o, its most advanced generative model yet. Shortly after, it disbanded a team dedicated to developing safeguards against "superintelligent" AI systems.
The team's disbanding made significant news. Reports indicate that OpenAI deprioritized safety research in favor of product launches like GPT-4o, leading to the resignation of the team’s co-leads, Jan Leike and OpenAI co-founder Ilya Sutskever.
While superintelligent AI remains largely theoretical, this week's coverage suggests that OpenAI’s leadership, particularly CEO Sam Altman, is increasingly prioritizing products over safety. Altman reportedly frustrated Sutskever by rushing AI-powered features for OpenAI’s first developer conference last November. He also clashed with Helen Toner, a former board member, over her critical paper on OpenAI’s safety approach, to the point of attempting to remove her from the board.
Over the past year, OpenAI has faced criticism for letting its chatbot store fill with spam and allegedly scraping data from YouTube against terms of service. The company has also voiced ambitions to let its AI generate explicit content. It appears that safety has taken a back seat, prompting some safety researchers at OpenAI to seek opportunities elsewhere.
Other Notable AI Stories
OpenAI + Reddit: OpenAI reached an agreement with Reddit to use its data for AI model training. While Wall Street welcomed the deal, Reddit users may have mixed feelings.
Google’s AI: At its annual I/O developer conference, Google unveiled numerous AI products, including video-generating Veo, AI-organized search results, and updates to Google’s Gemini chatbot apps.
Anthropic Hires Krieger: Mike Krieger, Instagram co-founder and recently co-founder of personalized news app Artifact, is joining Anthropic as its first chief product officer, overseeing consumer and enterprise efforts.
AI for Kids: Anthropic announced it would allow developers to create kid-focused apps using its AI models, provided they follow specific guidelines. Notably, competitors like Google do not permit their AI to be used in apps for younger audiences.
AI Film Festival: Runway held its second AI film festival, showcasing that the most impactful moments often come from the human elements rather than AI itself.
More Machine Learnings
AI safety is top of mind this week following OpenAI’s changes. Google Deepmind introduced a new “Frontier Safety Framework” to identify and mitigate potential harmful AI capabilities. This includes identifying dangerous capabilities, regularly evaluating models, and applying mitigation plans to prevent misuse.
Meanwhile, Cambridge researchers highlighted the ethical concerns of chatbots trained on deceased individuals' data. While potentially useful for grief management, this technology poses significant social and psychological risks.
On a less eerie note, MIT physicists are using machine learning to predict physical systems' phases, streamlining complex statistical tasks. At CU Boulder, researchers are exploring AI’s role in disaster management, focusing on improving resource allocation and training responders through human-centered AI approaches.
Finally, Disney Research developed a method to diversify outputs from diffusion image generation models, enhancing the variety and alignment of generated images.
Stay tuned for more updates in the fast-paced world of AI.