Anthropic scientists expose how AI actually ‘thinks’ — and discover it secretly plans ahead and sometimes lies

“We’ve created these AI systems with remarkable capabilities, but because of how they’re trained, we haven’t understood how those capabilities actually emerged,” said Joshua Batson, a researcher at Anthropic, in an exclusive interview with VentureBeat. “Inside the model, it’s just a bunch of numbers —matrix weights in the artificial neural network.” Credit: VentureBeat made with Midjourney

Tech leaders at Anthropic, IBM, and Meta warn that AI is coming for software developer jobs

In just three to six months, AI will be writing 90% of all the code produced, Amodei tells the Council on Foreign Relations. “In 12 months, we may be in a world where AI is writing essentially all of the code.” Programmers will still need to be around to dictate specific conditions, parameters, and goals, but he says that too may soon be outsourced to technology.
Chesnot/Getty Images

China’s Manus AI ‘agent’ could be our 1st glimpse at artificial general intelligence

Chinese startup Butterfly Effect has unveiled what it claims is the first general AI agent capable of acting autonomously. The AI agent, Manus, was developed by Chinese startup Butterfly Effect. Its representatives claim it is the world’s first general AI agent — meaning it demonstrates a level of autonomy that current AI models lack. The scientists who created Manus say it shows a potential glimpse of what artificial general intelligence (AGI) may one day be capable of. General AI agents promise to be more human-like than previous AI. (Image credit: Yuichiro Chino via Getty Images)

I put DeepSeek vs Meta AI Llama vs Qwen to the test locally on my PC — here’s what I recommend

AI giants like Google, Meta and OpenAI may grab all the headlines, but behind the scenes there’s an equally turbulent and rapidly expanding ecosystem taking shape. The market for small and specialized AI models that can run on home computers is reaching fever pitch, driven by a seemingly limitless demand for personal and business applications. (Image credit: Shutterstock)

The most underreported and important story in AI right now is that pure scaling has failed to produce AGI

The fact is, pure scaling has not worked. I am not alone in thinking this; the illustrious Stanford Natural Language Processing group reached a similar conclusion, reading between the lines of OpenAI’s recent announcement in the same way I did. In their words, Altman’s recent OpenAI roadmap was “the final admission that the 2023 strategy of OpenAI, Anthropic, etc. ‘“simply scaling up model size, data, compute, and dollars spent will get us to AGI/ASI’) is no longer working!” Jonathan Raa/NurPhoto via Getty Images