He explains that the human race has been focused on advanced analytics and artificial intelligence (AI) for many years, with the first papers on AI game theory published as far back as 1956. It then took nearly 40 years before a computer was able to take on the top chess players and win, as happened with Garry Kasparov against IBM's Deep Blue in 1997. However, he adds that this wasn't actually true AI, as much as it was a massive computer running the same algorithms, just much faster.
"After this, however, we saw things speed up significantly in this regard. By 2007, Chinook had solved ‘Checkers', while in 2011, IBM's Watson defeated Ken Jennings of ‘Jeopardy' fame. Then, between 2015 and 2017, AlphaGo defeated world-renowned ‘Go' players Fan Hui, Lee Sedol and Ke Jie. Following this, AlphaZero has dominated ‘Go', ‘Shogi' and ‘Chess'; while OpenAI has won online game Defence of the Ancients (DOTA) 2 in both single player and multiplayer modes," he says.
"One could say that in recent times, we are no longer in a position where such results can be considered to be ‘something that is 20 years away' to one where it is now a case of ‘unbelievable, it has already happened'."
Tullett points out that this rapid improvement in AI is built on the massive depth of investment that is being sunk into this technology. As a comparison, he indicates that in the past 20 years, Moore's Law has seen compute power increase approximately 12-fold, compared to the increase in AI power in the same period, which is more like a 300 000-fold increase.
"Of course, it's all very well pointing out how AI can defeat humans at various games, but what is it better at when we begin talking about business operations? For one, it is more effective at producing legal documents. In tests, neural networks produce a 94% accuracy, compared to 85% for the best performing humans. Moreover, where people took around an hour and a half to review these papers, the machine took just 26 seconds."
"Other areas where AI can prove to be a boon include para-legal research, transcription and translation, medical diagnoses - this is already beginning, and the expectation is it will lead to tremendous improvements in service delivery - and driving, since most automobile manufacturers consider self-driving vehicles to be a huge part of the future. Finally, AI is also good at interacting with people, as proven by the fact that Google Duplex has already beaten the Turing Test."
But what should be the key driver in enterprises adopting AI remains the fact that it is so good at beating humans at games, he says. After all, isn't business just another game, like Chess? Tullett suggests that business, like any other game, has a set of rules, it has a certain number of moving parts that need to be optimised by competitors in their pursuit of victory, and each competitor only has a limited knowledge of their opponent.
"Moving forward, business will need to embrace these technologies, but at the same time, they must neither expect miracles nor believe all promises. Nonetheless, there is the potential for enterprises to begin solving today's problems with tomorrow's technology. Of course, such a strategy must incorporate strong security, since the bad guys also use AI, while internally, companies need to develop flexible skills that are not silo-specific, so they can more easily identify and attack specialised opportunities."
"You may not think your organisation needs to dive into AI just yet, but heed this warning: You can no longer expect years to go by before you face a competitive challenge from rivals using AI to help drive their business forward. By 2020, it is anticipated that some 20% of large enterprises will already be utilising machine learning software to create business materials, such as company reports. This leaves you to answer just one final question: once you know that your competitors already plan to have these technologies in place, can you really afford not to be doing the same?"View more content