What I learnt from talking to Google about chatbots


Those attending the World Economic Forum earlier this month had a plethora of solemn issues to discuss: the energy crisis, war in Ukraine, China-US tensions and inflation, to name but a few. But the topic creating a buzz over the dining tables was something else: ChatGPT, the artificial-intelligence platform recently launched by OpenAI, a small American start-up, that appears to converse and create content (almost) like a human. 

“It’s all anyone wants to talk about,” one chief executive officer ruefully remarked, as the Davos attendees swapped tales about their own experiments with ChatGPT in the office, and fretted over whether their kids were using it to do homework.

The fascination partly arises because its potent capabilities make it “a game-changer that society and industry need to be ready for”, as the WEF says. But it also points to an intriguing question: could this so-called generative AI, which answers questions, undermine the dominance of the mighty Google Search platform? Is a Silicon Valley giant, infamous for disruption, about to be disrupted itself? 

Until recently it was hard to imagine that anything might challenge Google’s dominance in search. In the quarter century since the tool was first created it has embedded itself in most of our lives as well as our language, even becoming a verb in its own right. Google accounts for 85 per cent of the market, with 99,000 Google searches taking place each second (a mind-boggling 8.5bn each day). Next to this, rival platforms such as Microsoft’s Bing or Yahoo are fighting over the scraps.

The figures would seem to put Google in an unassailable position. Regulators have periodically fretted about its market dominance. They also explain why Google’s parent, Alphabet, is now worth $1.3tn. But, like David and Goliath, sometimes the little guy can surprise a giant by attacking in a novel way. And some of the Davos crowd think that ChatGPT could end up doing precisely that. 

Some at Google think it makes sense if a tiddler such as ChatGPT acts as a lightning rod for the early public angst around generative AI

Certainly it’s not too hard to imagine that a slick AI-enabled tool could offer a more convenient way to search for information in the future. Particularly since Microsoft has just taken a $10bn stake in OpenAI, and plans to integrate the system into its products to “completely transform” them, as CEO Satya Nadella says. What’s notable about this innovation is that it did not come from one of the tech giants at all. On the contrary, Google has so far failed to deliver anything comparable to the market, despite historically priding itself on its artificial-intelligence capabilities.

In search of answers, I broached that question with some top executives at Google, who insisted that the issue revolved around ethical matters. The Alphabet leadership claims the company has developed AI platforms more powerful than ChatGPT but has decided it cannot release them until it has found a way to control the potential social and ethical risks. “Our aim is to be bold and responsible,” says James Manyika, head of tech and society at Google. “We believe Google is very competitive in this space.” 

A different executive added that “being the first mover is not always a good thing”. Indeed, some at Google think it makes sense if a tiddler such as ChatGPT acts as a lightning rod for the early public angst around generative AI, since it will make it easier for bigger companies to move into the space later.

Perhaps. Tech history is full of tales of “second mover” companies that have emulated an innovation that was unleashed by somebody else, and then eclipsed the pioneer. MySpace and Facebook is a case in point. However, some executives at Google’s rivals – the likes of Microsoft and Amazon – offered another explanation: they think the company is haunted by the problem of incumbency, as described by the US business consultant Clayton Christensen in his 1997 classic The Innovator’s Dilemma. Perhaps, they suggest, it’s because Google Search is so dominant that staff have no incentive to scramble to develop a technology that might undercut its past success.

There is another issue I wrote a book about: silos. OpenAI has a single brainstorming group that is developing its innovation. Alphabet, however, has one team inside Google which is working on AI, and another at the DeepMind AI company it acquired.

On paper, that looks like double the talent. But history is also full of companies that failed to innovate because different departments competed rather than collaborated, ultimately undermining the scope for innovation. Sony has suffered from this problem before. So too has Microsoft.

Alphabet is formidable enough to overcome all these issues. But the excitement over ChatGPT shows why innovators, and especially incumbents, can never afford to relax.

Follow Gillian on Twitter @gilliantett and email her at gillian.tett@ft.com

Follow @FTMag on Twitter to find out about our latest stories first

Leave A Reply

Your email address will not be published.