>
Google subsidiary DeepMind says it could be launching a ChatGPT rival soon — and its chatbot promises to be a safer kind of AI assistant.
DeepMind has been a pioneer in AI research for the past decade and was acquired by Google nine years ago. However, with ChatGPT stealing recent headlines, DeepMind CEO Demis Hassabis told me Time (opens in new tab) that it is considering releasing its own chatbot, called Sparrow, for a “private beta” sometime in 2023.
Sparrow was introduced to the world last year as a proof-of-concept in a research paper (opens in new tab) that described it as a “dialogue agent that is useful and reduces the risk of unsafe and inappropriate answers”.
Despite some doubts about the potential dangers of chatbots, which DeepMind says contain “inaccurate or fabricated information,” it seems Sparrow could soon be ready to fly in beta form. In fact, given DeepMind’s close relationship with Google, it could become the search giant’s answer to ChatGPT.
The slight delay in Sparrow’s launch, according to Demis Hassabis, is due to DeepMind’s willingness to ensure it has important features that ChatGPT lacks, particularly by citing specific sources. As Hassabis told Time, “it’s good to be careful on that front.”
From DeepMind’s research paper it also sounds like Sparrow will initially be more restricted and conservative than ChatGPT. The latter has gone viral with its impressive ability to help everyone from programmers to armchair poets, but it also caused alarm with its capacity for discriminatory comments and malware writing skills.
DeepMind has discussed the behavior-restricting rules Sparrow has built upon, along with his willingness to refuse to answer questions in “contexts where it’s appropriate to defer to people.” In early tests, Sparrow apparently gave a plausible answer and, crucially, backed it up with evidence “78% of the time when a factual question was asked”.
But the true possibilities will become clearer when that public beta launches later this year. We’ll be sure to grab the popcorn for the first AI chatbot debate between the Google-affiliated Sparrow and the increasingly Microsoft-loving ChatGPT.
Analysis: AI chatbots are still in kindergarten
Anyone who has used ChatGPT knows that it is capable of mimicking intelligence on a number of subjects quite well. But while that’s certainly a fun ride, AI chatbots also need moral intelligence and the ability to cite sources – which is where DeepMind says its Sparrow “dialogue agent” is strongest.
Taking this to the next level requires a ton of external input, which is why a public beta of Sparrow is about to drop. DeepMind says developing better rules for its AI assistant “will require both expert input on many topics (including policymakers, social scientists and ethicists) and participatory input from a wide range of users and affected groups.”
Sam Altman, CEO of OpenAI (which created ChatGPT), has similarly spoken about difficulties in opening AI chatbots without causing collateral damage. on Twitter (opens in new tab) he admitted, “there will be significant problems over time using OpenAI technology; we will do our best, but will not successfully anticipate every problem.”
In other words, the developers of both ChatGPT and DeepMind’s Sparrow are like parents with curious toddlers, bringing both fun and danger – especially when their kindergarten teacher is, in fact, the entire Internet.
ChatGPT is already wild and moving towards a monetized future with ChatGPT Professional, an upcoming paid tier. But DeepMind’s Sparrow sounds like it could be the more gentle nature AI chatbots need as they race to next-gen models like the rumored ChatGPT-4.