Originally posted on venturebeat.
Back in April, Google Brain and DeepMind joined forces as Google DeepMind, with plans to take on the competitive threat posed by OpenAI and its game-changing ChatGPT. Now, according to Wired, DeepMind CEO Demis Hassabis says the company is working on a new system, Gemini — which was teased at Google I/O in May — that will “tap techniques that helped AlphaGo defeat a Go champion in 2016.”
According to the report, Hassabis said the Gemini system will combine LLM technology with reinforcement learning techniques used in AlphaGo, with a goal of giving it new planning and problem-solving capabilities.
Another recent report from The Information cited an anonymous source who said that Google’s researchers have been using YouTube (which Google owns) to develop Gemini — which AI practitioners say could be an edge for GoogleDeepMind, since it can get “more complete access to the video data than rivals that scrape the videos.”
At Google I/O, CEO Sundar Pichai noted in a blog post that Gemini, the company’s “next-generation foundation model,” is still in training.
“Gemini was created from the ground up to be multimodal, highly efficient at tool and API integrations and built to enable future innovations, like memory and planning,” he wrote. “While still early, we’re already seeing impressive multimodal capabilities not seen in prior models. Once fine-tuned and rigorously tested for safety, Gemini will be available at various sizes and capabilities.”
The reports about Gemini come as Hassabis himself has expressed concern about the long-term, existential risks of AI. In the Wired report Hassabis says that no one really knows for sure that AI will become a major danger, but that if progress continues at its current pace, there isn’t much time to develop safeguards. “I can see the kinds of things we’re building into the Gemini series right, and we have no reason to believe that they won’t work,” he said.
But Hassabis also recently was one of the most prominent signatories on the Statement on AI Risk, which warned against a “risk of extinction” from advanced AI if its development is not properly managed.