This photo illustration shows the ChatGPT logo at an office in Washington, DC, on March 15, 2023.
Stefani Reynolds | AFP | Getty Images
We are still some way off reaching human-level artificial intelligence despite rapid advances in the technology, according to an early investor in research laboratory DeepMind.
“In terms of artificial general intelligence, OpenAI, ChatGPT stuff: it’s like saying we’re going to jump to the moon,” Humayun Sheikh, a founding investor in AI startup DeepMind, which is now owned by Google, told CNBC in an interview.
related investing news
“We took a big jump, but we’re not at the moon yet.”
Sheikh, who held around 1.3% of DeepMind’s shares in 2011, said that large language models (LLM) like those developed by Microsoft-backed firm OpenAI, though impressive, are lightyears away from so-called artificial general intelligence, or AGI.
AGI is often referred to as the holy grail of AI. It is a hypothetical system capable of completing any task to the same level as a human.
This is very much: Google is not born yet, but Yahoo is.
Humayun Sheikh
CEO, Fetch.ai
“That’s really how I compare AGI with all the large language model companies which are popping up,” Humayun, who is now co-founder of AI and blockchain startup Fetch.ai, said.
“They are very limited. How you actually get them to do certain things is still in its infancy.”
“This is very much: Google is not born yet, but Yahoo is,” he added.
His comments come as Google-parent Alphabet merges DeepMind with Google Brain, part of the U.S. internet giant’s research division.
Google is racing to compete with Microsoft and other tech companies in the field of AI. Microsoft is making huge strides with its investment into OpenAI and the inclusion of the firm’s LLM technology into its Bing browser and other products.
Earlier this week, Demis Hassabis, the founder of DeepMind, told the Wall Street Journal that some form of AGI might be possible “in the next few years.”
Sheikh said he had a “lot of respect” for Hassabis and that the entrepreneur is “very aware of the ethics of AI.”
“One of the first things he has been always bringing out was, how do we control it? How do we put that boundary around it and make sure AI doesn’t go out of control?,” Sheikh said.
Google acquired DeepMind for $500 million in 2014 and is attempting to bolster its business by doubling down on AI in a bid to fend off the threat to its core search unit from OpenAI. Google launched its own chatbot alternative to ChatGPT, Bard, in March.
AI’s huge potential is seen in its ability to generate entirely new content from user prompts. People have used the technology to create everything from poems to quirky images and movie trailers, while kids are using it to help with their homework.
Experts have raised concern over the risks of sophisticated AI, however, with a group of tech leaders including Elon Musk and Steve Wozniak calling for a six-month ban on the development of AI more advanced than GPT-4, the latest version of OpenAI’s massive language processing software.
WATCH: Can China’s ChatGPT clones give it an edge over the U.S. in an A.I. arms race?
