DeepMind, a Google company, has hinted that it may soon release a ChatGPT competitor; this new chatbot would provide a more secure artificial intelligence helper.
Google purchased DeepMind, a leader in artificial intelligence (AI) development, nine years ago. Despite ChatGPT’s recent success, DeepMind’s CEO Demis Hassabis recently told Time that the company is thinking about launching its own chatbot, Sparrow, for a “private beta” in 2023.
A proof-of-concept for Sparrow was published in a scientific journal last year, introducing it to the public as a “conversation agent that is beneficial and decreases the danger of harmful and unsuitable replies.”
Despite concerns about the risks of chatbots, which DeepMind claims include “inaccurate or created information,” it seems that Sparrow might soon be ready to take flight in beta form. Due to its tight ties with Google, DeepMind has the potential to replace ChatGPT as the company’s primary messaging platform.
Demis Hassabis has said that the delay in the release of Sparrow is due to DeepMind’s eagerness to ensure that it contains critical functionality that ChatGPT lacks, most notably, citing particular sources. Hassabis agreed with Time’s assessment that “caution is warranted” on that front.
Based on DeepMind’s white paper, it seems that Sparrow will launch with fewer freedoms and more conservatism than ChatGPT. Even while the latter’s capability for discriminating remarks and malware-writing abilities has raised anxiety, it has gone viral thanks to its amazing ability to aid everyone from programmers to armchair poets.
DeepMind has lauded Sparrow’s ability to refrain from answering queries in “contexts where it is acceptable to defer to humans” and its adherence to behavior-restraining constraints. Early testing showed that “78% of the time when posed a factual question,” Sparrow delivered a credible response and importantly backed it with evidence.
The actual extent of its capabilities, though, won’t be known until the public beta is released later this year. Definitely going to get the popcorn for the first AI chatbot discussion between Sparrow (which is linked with Google) and ChatGPT (which is becoming more loyal to Microsoft).
AI chatbots have not yet graduated from preschool
Any user of ChatGPT can attest to the fact that it can convincingly pass for intelligent on a variety of topics. Though that’s exciting, DeepMind claims that its Sparrow ‘dialogue agent’ excels in areas such as moral intelligence and the capacity to cite sources in conversations.
Because we need so much feedback from the public to take Sparrow to the next level, we’re about to launch a public beta. DeepMind claims that it “will need both expert input on many issues (including politicians, social scientists, and ethicists) and participatory input from a broad variety of users and impacted groups” in order to establish better guidelines for its AI assistant.
The CEO of OpenAI, the company responsible for developing ChatGPT, Sam Altman, has also discussed the challenges of releasing AI chatbots without creating collateral harm. ‘There are going to be big challenges with the adoption of OpenAI tech over time,’ he said on Twitter (opens in new tab), ‘we will do our best but not effectively predict every issue.’
What this means is that the creators of ChatGPT and DeepMind’s Sparrow are analogous to parents of curious toddlers, with all the joy and peril it entails—especially when their kindergarten instructor is essentially the whole internet.
Since ChatGPT is taking off like a rocket, a premium paid version, ChatGPT Professional, will soon be available. Sparrow from DeepMind, on the other hand, may be the more mellow personality that AI chatbots require as they rush toward the next-generation models like the reported ChatGPT-4.
Subtly charming pop culture geek. Amateur analyst. Freelance tv buff. Coffee lover