Google's search for an AI future as it turns 25

The tech giant Google and I almost share the same birthday... give or take a few years.

Google turns 25 this month (I'll have a few more candles on my cake) - and finds itself in a tech landscape that has changed dramatically since founders Larry Page and Sergey Brin started it in 1998.

Back then Google was only a search engine, and it lived for its first few months in the garage of Susan Wojcicki - the future boss of YouTube.

You do not need me to tell you how well that search engine worked out. It has been 17 years since the word Google officially entered the dictionary. I remember a BBC discussion about whether we should use it as a verb on-air because of its potential to be a free advert for the firm.

That company - now part of a larger parent group called Alphabet - has since diversified into pretty much every area of tech and dominates some of them to an extent which sometimes troubles anti-competition regulators. Right now it is trying to Google itself into pole position in the AI race - but some say it has already fallen behind.

Hits and misses

Email and smartphones, software and hardware, driverless cars, digital assistants, YouTube - Google has spawned (and acquired) hundreds of products and services. Not all of them have worked out.

There are 288 retired projects listed on the Killed by Google website, include gaming platform Stadia and budget VR headset Google Cardboard.

Google vice-president Phil Harrison showed off the Stadia controller on-stage at its launch in 2019

The question now is whether Google can maintain its omnipresence in the rapidly evolving world of artificial intelligence.

There have been mutterings, including from within, that it has fallen behind. A leaked memo from a Google engineer found its way on to the net, in which he said the firm had no AI "secret sauce" and was not in a position to win the race.

This feeling was further fuelled by the battle of the bots.

• What is AI, is it dangerous and what jobs are at risk?

• 'Google killer' ChatGPT sparks AI chatbot race

• Google what our chatbot tells you... says Google

For many people, the first time they knowingly interacted with AI - and were impressed by it - came in the form of ChatGPT, the viral AI chatbot which exploded into the world in November 2022.

Its creator OpenAI has received billions of dollars in investment from Microsoft, which is now working it into its own products, including the Bing search engine and Office 365.

ChatGPT has been dubbed the "Google killer" because of the way it can answer a question in one go, rather than serve up pages and pages of search results.

It uses a language-processing architecture called a transformer which was actually invented by Google, but when Google followed up a few months later with its own rival Bard, it had nowhere near the same impact.

Bard was given a surprisingly cautious launch. It was not for under-18s, the tech giant said, and it was described to me as "an experiment" by a senior exec.

Perhaps part of its caution was in part a result of a weird situation which preceded Bard.

Sentient debates

The foundation of a chatbot is a Large Language Model, or LLM. One of Google's original LLMs was called Lamda.

An engineer who worked on it became convinced that it was sentient. He published pages and pages of conversations which he claimed proved his point - that Lamda was sharing real emotions and thoughts.

This is of course exactly what an LLM is trained to do - generate authentically human-like text. Google has consistently denied that Lamda was doing anything more than that, and the engineer was fired.

But it created headlines around the world and heightened nervousness about AI long before the debate was topical - headlines of which perhaps Google would have preferred not to have been part.

The firm has certainly not given up. At the IO developers conference in May, Google announced 25 new AI-driven products. In a statement on its website it declares itself to be "at the forefront of advancing the frontier of AI". It owns leading UK-based AI firm DeepMind, whose AI program AlphaFold has the potential to supercharge the discovery of new medications.

In August, Chirag Dekate, from the analyst firm Gartner, attended the future-gazing event Google Next, which he said was all about AI.

"Google is gearing for leadership in the emerging Generative AI economy," he says.

'Not-so-secret weapon'

Analyst Carolina Milanesi, from the firm Creative Strategies, also thinks we should not be so quick to write off Google AI.

"I do not really buy into the 'they missed the boat on AI' conversation," she says.

"The opportunity for them is in AI both on the consumer and enterprise side."

Susannah Streeter, head of money and markets at the investor Hargreaves Lansdown, agrees.

Google's not-so-secret AI weapon could lie in its successful cloud business, she argues. Cloud companies offer access to huge networks of computers and processing power that would be logistically difficult for most companies to own or house by themselves.

"Alphabet is positioning itself to be at the centre of the AI revolution with its Google Cloud business, given the huge demand from businesses large and small to update infrastructure and storage ready to whirr through huge generative AI workloads," she says.

"It may still be the smallest of the three big cloud providers, behind Amazon Web Services and Microsoft Azure - but it still packs a punch."

Journalist Tim Dowling once wrote about avoiding Google's consumer services for one week. He described trying to organise a trip to the cinema as being "like shopping by candlelight".

If just a fraction of Google's AI products can embed themselves in the same way, it should be able to keep the lights on.