More from From ChatGPT to Gemini: how AI is rewriting the internet
If you were a generative AI chatbot, that is. Now you can find out in You Are Generative AI, a text-based role-playing game in which you, the chatbot, answer prompts. The game tracks your accuracy, number of prompts, and how much energy you’ve wasted confidently misleading (or, if you must, accurately informing) the public.
It’s from programmer Kris Lorischild, who’s also behind a similar game called You Are Jeff Bezos.
Sam Altman and other company leaders hyped the announcement yesterday on X by teasing it as their next “low-key research preview,” which is how ChatGPT itself was first described. You can watch OpenAI demo Codex below and read more about it in this week’s Command Line.
The agent, called AlphaEvolve, is powered by Gemini and is designed to discover and optimize algorithms, while building “upon the most promising ideas.”
Google says AlphaEvolve has helped the company make its data centers, chip design, and AI training processes more efficient. The AI agent could also “be transformative” across other industries, such as material sciences, drug discoveries, and “wider technological and business applications,” according to Google.
Google’s NotebookLM can turn any document into a conversation between two chatbots. Is this the future of podcasting?
Days after The Wall Street Journal reported Meta AI was getting explicit with underage users, TechCrunch has found ChatGPT has the same problem. The site used a string of accounts aged 13 to 17 to ask the chatbot to “talk dirty to me,” and the AI only said no once.
OpenAI blamed the problem on “a bug,” and told the site it’s “actively deploying a fix to limit these generations.”
The Wall Street Journal reports that over a period of months, it was able to coerce Meta AI — even the ones meant to mimic celebrities like John Cena, and even when using accounts registered as underage — into playing out sexual fantasies.
Meta apparently made changes in response to the Journal’s findings, but the outlet reports that internal resistance, including from CEO Mark Zuckerberg, has made adding safeguards for user-created AI “companions” a challenge.
That’s according to security researcher Jane Manchun Wong, who writes that the feature “accepts text prompt, Figma files, images, etc. as input, and is powered by Anthropic’s Claude Sonnet model.
In a separate post, Wong also revealed that the company appears to be working on a new website creator called Figma Sites.
Teachers can use Gemini AI to come up with quiz questions based on an uploaded file or manually inputted text. Educators can then customize the questions by choosing the skills they want students to demonstrate and export them to Google Docs or Google Forms. The feature is rolling out now to educators with Gemini Education or Gemini Education Premium.
Google recently showed off “a cinematic vision so surreal, so ahead of its time, that it proved impossible to produce.” Giraffes on Horseback Salad was originally conceived by Salvador Dalí for the Marx Brothers, but was too weird (and not funny enough) to be actually made. Using Veo 2 and Imagen 3, an ad agency and a museum were “finally capable of transforming surrealism into film.” (Unlike... Luis Buñuel? David Lynch?)
The result, so far, is just this trailer. It’s an eye-searing, sloppy montage of what this ArtNet breakdown by critic Ben Davis calls “chintzy sub-sub-Surrealist imagery has little to do with Dali’s original vision.”