30s Summary

Dario Amodei, CEO of Anthropic, predicts human-level artificial intelligence (AI) could be reached as early as 2026, equating the evolution of AI to academic progression. He notes potential challenges such as data limits, scalability, and political interference. The potential power of AI was also discussed, and he encouraged responsibility with its use. The company’s AI chatbot, Claude, competes well with rivals like Open AI’s ChatGPT, but no release date for the next version was given. The claim of near-term artificial general intelligence (AGI) echoes predictions made by OpenAI CEO Sam Altman.

Full Article

Artificial intelligence is sprinting up the academic ladder, and could be on par with human intelligence by 2026 according to Dario Amodei, Anthropic’s CEO. Just recently, he was chatting with Lex Fridman on his podcast where he likened the progress of artificial general intelligence (AGI) to an academic journey.

“Imagine it like this: last year we were at the undergrad level, the year before, we were as smart as high school students, and now we’re nearing the completion of our PhD.,” Dario quipped. He also cheekily predicted that AI might outpace humans in just a few short years, “If you just squint your eyes and look at the pace we’re going, I’m thinking we’re hitting human levels by 2026 or 2027.”

Of course, he cautioned that numerous things could trip up the progress, like running out of data, a limit to scaling, or even political issues messing with microchip production. But assuming all goes well, he repeated, “we’re on track for 2026 or 2027”. However, he did say there might be a slight delay.

Dario made a point about the power of AI too, saying, “Like with all powerful things, they can do good and bad,” and adding, “With great power comes great responsibility.”

Anthropic is very proud of its chatbot Claude, who stands up strong to rivals like OpenAI’s ChatGPT. When faced with questions about the competition, Dario reiterated the company’s mission to keep pushing boundaries and inspiring others to do likewise.

Also, when Lex asked when the company’s AI chatbot, Claude Opus 3.5, was expected to launch, Dario remained mum about a specific date.

Sam Altman, OpenAI’s CEO, has also made similar claims about AGI arriving soon. He’s quite confident that his company will achieve AGI within the next five years with current hardware.

On a more academic note, AGI is basically a type of artificial intelligence that can match or even exceed human capabilities on any cognitive task. It’s like the super-smart robot in your sci-fi dreams.