In order to advance deep learning and artificial intelligence, companies need to not only open up their data for research, but also the problems that create roadblocks for their business.
This is according to Daniel Hulme, CEO of Satalia – a company that uses advances optimisation algorithms to solve problems for businesses, who was speaking yesterday at the RE WORK Deep Learning Summit in London.
“It’s ok for companies to open up their data, but they also need to open up their problems,” he said.
In doing so, he said they would be inviting people to solve their problems for them, in a similar way to white hat hackers’ practice of identifying security loopholes and alerting the appropriate company.
This would enable faster innovation, as problems would – at least in theory – be more rapidly solved, allowing newer and better products to be produced more quickly.
However, while an excellent idea in theory, Hulme acknowledged that in reality there were some significant issues with this approach, primarily due to a lack of willingness on companies’ parts to share such information.
“Companies don’t want to do that,” he said, pointing to concerns over embarrassment and a desire by companies to be seen as perfect.
There is, of course, the additional issue of shareholder confidence – if a company were to admit it was having problems optimising a major part of its business, it may well have a significant effect on share price.
Nevertheless, for startups in the field of deep learning and AI, problem solving is the bread and butter work that enables them to become fully-fledged businesses.
“I think when you’re a small company – a startup – you need to solve problems for people,” he said.
Hulme added that while some startups took a different approach, by pursuing a blue-sky concept without a specific marketable product in mind, this required a massive amount of funding and so was only feasible if the company was founded by someone with significant personal wealth.