Already a subscriber? Log in
Other computer scientists were focused on teaching machines the rules of language, the theory being that if you told a computer everything about grammar, it would be able to understand humans. ‘It hasn’t been very successful,’ Lynch explains. ‘If I said “the dog walked into the room, it was furry,” there’s no rule of language that tells you what the “it” refers to. You use your knowledge and assume it’s the dog; there might be a situation in a 1970s Manhattan loft where there’s a furry room, but it’s highly unlikely.’
‘My solution was to do something very counterintuitive, and treat the problem as a mathematical one, something which makes every English teacher draw a sharp intake of breath,’ says Lynch. He set about developing software which could spot bias in seemingly random processes, such as human conversations; the chances of seeing the word ‘dog’ next to ‘bark’ was greater than it would be in other instances, and — as long as it read enough content — a computer could eventually work this out.
‘The grammatical method is like being 11 and studying French from a textbook, while learning the language mathematically is like being two-and-a-half and growing up in Paris.’
Lynch is excellent at simplifying complex technological ideas by using analogies and examples. It is easy to imagine him as the best sort of university lecturer — but as the importance of his work began to emerge, it was clear he would leave academe.
‘The excitement in the Cambridge days was always about solving a problem. You don’t think about what you might actually do with it,’ he says. What he did, in 1996, was to found Autonomy. He has since transformed it into a FTSE 100 firm, with 16,000 major clients around the world ranging from Nestlé and the BBC to NASA and the US Department of Homeland Security. Autonomy recently reported record first-quarter results, with a pre-tax profit of $58 million, 87 per cent up on the same quarter last year.
As you might expect, Lynch approached his business plan from a mathematical perspective, asking which problems were the most valuable. Unsurprisingly, it turned out to be those belonging to big corporations, espec- ially financial institutions, which were struggling to make sense of new obligations relating to the vast amount of data they hold. ‘Lots of the major banks use the system to read staff emails, which now have to be kept and monitored by law, and to listen to phone calls from the trading floor. The system can understand what’s going on and spot when people are committing a fraud,’ he explains.
Autonomy also helps firms prepare for litigation and the onslaught of new regulation related to the financial crisis. The Financial Services Authority and the US Securities & Exchange Commission are two of its biggest clients. It was, says Lynch, a conscious decision to develop a product that would thrive in bad times.
‘Having lived through the dotcom bubble, you realise that technology is just like farming. If there is summer, there will be winter, so we took the decision to work out what would work in winter. In the current climate, regulation and litigation are holding up very well.’
But his technology’s ability to let machines understand humans has far broader applications. ‘We’re used in call centres, where the software listens to the conversations and can alert a supervisor when a customer is getting agitated. It can also listen to the customer’s question and put up the correct answers. A call-centre rep might not be the world’s brightest person, which is why you rarely get the right answer. But in Vodafone’s centres, we’ve reduced by 30 per cent the amount of time it takes the rep to give the correct advice.’
One of the advantages of using a maths-based approach is that there’s little the software can’t understand. It is equally comfortable comprehending quantum physics as it is a call-centre conversation — one of Autonomy’s 900 scientists even taught it Klingon, the Star Trek language.
It’s here that the future of the technology becomes so exciting. ‘The most famous example we have is an aerospace manufacturer. It bought our system and, on day one, the system spotted similarities in two projects and said, “Hey, you guys need to talk to each other.” That saved the firm $7 million because they were working on the same problem.’
If it can understand aerospace technology, there’s no reason it can’t trawl through all the research on cancer, spotting unseen links, or monitor suspected terrorists and alert intelligence services to a planned attack.
Lynch is adamant that his business is not about replacing people, however. ‘A well-educated human being is very difficult to beat at anything — until the first 25 minutes have gone by and the boredom sets in,’ he explains. Much of what Autonomy does involves sifting through reams of data and presenting the most relevant pieces to a real person. The techno- logy is also in its early stages, and finds it hard to deal with the subtleties of human inter-action; it wouldn’t spot that George Orwell’s Animal Farm is an allegory, for example, and finds it difficult to deal with sarcasm.
The firm’s link with Cambridge University remains strong — and the university still produces the best graduates in the world, says Lynch. But he admits that founding a tech firm in the UK hasn’t been easy, often because the British struggle to sell themselves. ‘When students from Cambridge and the Massachusetts Institute of Technology pitch against each other at exhibitions, I’m often asked who I’d invest in. I’d always take ideas from Cambridge, but get the MIT lot to sell them.’ In a sense, Lynch is proof of this; born in Ireland but brought up in Essex, the son of a fireman, he is articulate and affable — but also very understated and oh-so-English, constantly fiddling with his wedding ring and peering across to read my notes. Although he’s been dubbed ‘Britain’s Bill Gates’, it’s hard to imagine him struttin g a huge stage with a microphone.
More worryingly, Lynch says the best British computer scientists are leaving for San Francisco and Silicon Valley, largely because our society and tax system don’t reward entrepreneurs. ‘The saddest thing is that whenever I’m in California, I can go out every night with Brits. It’s important to say that the early years of the Blair government were good for entrepreneurs, and there were a lot of things done that sent messages about how we created value for everyone, not just the few.’
Since Gordon Brown’s arrival, much of that work has been undone, Lynch says. He cites a loophole that let businesses that were still being run by a founder pay a lower rate of capital gains tax. It was closed, largely because it was being exploited by the private equity industry. ‘The message that someone who remortgages their house and stays up all night every night for three years to found a business is the same as a private-equity fat-cat is ridiculous.’
As I leave Lynch’s office, the neon lights continuing to buzz Bayesian equations, my mind turns to the Presbyterian minister who departed this life in 1761 and was buried in Bunhill Fields cemetery in London. A shy, modest man, he never published his work, and did not succeed in proving what he most dearly believed. Now, two and a half centuries later, it has an application far beyond his wildest dreams. I wonder if he knows and, if he does, what on earth he thinks of it all.
Also in Society
Latest
Comments
Join the debate for just $5 for 3 months
Be part of the conversation with other Spectator readers by getting your first three months for $5.
UNLOCK ACCESS Just $5 for 3 monthsAlready a subscriber? Log in