Late on Monday an announcement went up on Google’s company blog in the form of “a message” from Sundar Pichai, CEO of Google and parent company Alphabet, formally introducing the world to its artificial intelligence (AI) chatbot, Bard.
It was not an unexpected announcement, having been signposted during the latest Google earnings call last week. Still, the lengthy update has set tech tongues wagging that “Google’s answer to ChatGPT” puts it back on top of the AI bunfight.
“ChatGPT’s complete dominance in the world of AI-powered conversational platforms is at an end,” proclaimed an article on LifeWire, for example.
But it is far too soon to crown a new champ, as Bard — we were told in the blog — is now being used by “trusted testers” and will only be available for public usage “in coming weeks”. Its viral moment of stunning millions of users and stealing headlines is yet to come.
Furthermore, if futurists and AI researchers are to be believed, we are at the start of a sharp upward curve of AI capability and mainstreaming, a trajectory towards something like artificial general intelligence (AGI) and the singularity.
So, what can we expect from Bard? Pichai calls Bard an “experimental conversational AI service” (aka a chatbot), powered by Google’s language model for dialogue applications (LaMDA), which was unveiled two years or so ago.
Google says Bard will bring together “the breadth of the world’s knowledge” (read: all the indexed web Google already navigates and relates so well, and more) with “the power, intelligence and creativity of our large language models”, to parse and present that information in useful and creative forms.
The examples (plan a party, plan a meal, explain something) show how complex queries can prompt Bard to offer human-readable information that is both up-to-date and tailored to task.
The Bard platform will initially use a “lightweight” iteration of LaMDA as this requires less computing power, so it can be used at scale, which means not falling over when every excited techie, bored professional and cunning student descends on it, hoping to compare it to the generative capabilities of GPT3.
Pichai also promised that we will soon see AI-powered features embedded into the current Google search function. AI is, of course, already at play within search. That’s how it is able to offer you results like “key moments in videos”. But now we can expect to reap the benefits of its various AI projects (LaMDA, MusicLM, PaLM and other acronyms, as well) that are all about novel models for processing multimedia information.
This announcement has come at a critical time for the company, which has been beset by bad news of late.
In practical terms, if I type “tell me about Google’s Bard AI” into Google search now, I get a list of related web pages based on keywords, as well as a “featured snippet” (an extract from a page that Google’s system has decided is likely to be most relevant). There’s also list of related queries under the heading “People also ask”. We’ve come a long way from Google’s early page rank system.
Those results are a great start to understanding Bard, but still not generative. It’s not fulfilling the “tell me” part. Bard will change that, using large language and diffusion model capabilities in combination with search to produce a new answer synthesised from that pool of information resources. Language models can also be used for translation, summaries and comparisons, and in this way provide replies to the kind of more complex query Google users are already asking of it.
In short, deploying its new AI tools means the Google search function will better understand what is being asked (the nuance and specificity of natural language) and serve users more relevant “synthesised insights for questions where there’s no one right answer”.
A generative language API will also become available to Google clients (“companies, developers and creators”) from March, so developers can use the language models within their own applications.
This announcement has come at a critical time for the company, which has been beset by bad news of late. Alphabet’s fourth quarter 2022 results show the depressive effect of the economic downturn on search advertising — more than a glancing blow to the company’s core moneymaker. As Bloomberg reports, this feeds into “investor anxiety that wildly popular AI rivals, like Open AI’s ChatGPT, could upend search”.
Google has also announced layoffs recently, and there has been the blow to its reputation as the leader in cutting edge technologies, including AI tools.
There are moments in the blog that show Google’s keenness to claim back that space, and its title within it. Highlighting that Bard will be connected to the web, and thus able to create “fresh” responses, “ground[ed] in real-world information” feels like a not-entirely-oblique reference to ChatGPT’s acknowledged shortcomings.
In this case, the fight has come to Google. Software giant Microsoft is a major investor in ChatGPT, and insiders have said it is integrating ChatGPT into Bing, which could give its lacklustre search rival just the boost it needs to be a more serious challenger to Google.
In fact, at the time of writing on February 7, analysts and media are waiting to hear what Microsoft has up its sleeve after the company said on Monday (shortly after the Google blog went up) that it would be holding an in-person event at headquarters on Tuesday evening to “share some progress on a few exciting projects”.
That’s not much to go on, but many have been expecting this to be confirmation of the ChatGPT integration. Either way, the game’s officially on.
• Thompson Davy, a freelance journalist, is an impactAFRICA fellow and WanaData member.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.