What's fresh

June Question of the Month: When do you think that AI will outperform humans at work?

AI has had a lot of hype around it this year, from the breakthrough in driverless cars being eyed up by Uber, to artificial intelligence being used to help judge 590,000 submissions of art at the EyeEm Photography Awards. Even the ancient Chinese game of Go, that for 2500 years has always had a human best player, was won by Google's AlphaGo deep reinforcement learning AI program when playing against the reigning Go world champion Lee Sedoul.  

So this month we wanted to look to the future and see what the new advancements in machine learning mean for different businesses.

Our results showed that the year 2032 will definitely be one to watch out for as the majority of people believed that within the next 15 years AI would outperform humans across all sectors. Within Financial Services, Consulting and Marketing most people think this revolution will take 11-15 years but in Retail and HR, the change may happen even quicker, with 55% saying as soon as 2022. In many ways the Retail industry has seen a major change with the growth of AI already with self-checkouts and online shopping making humans less and less needed in this sector and causing the demise of 'brick and mortar' stores as we have analysed previously [Read our article on 'uniting the online and offline experience' article]. If driverless cars become mainstream soon, then logistics for deliveries will also not need a human to control them. 

However, although in HR our results showed this AI revolution might be happening quickly, 82% believed that the Financial Services sector will have an AI revolution within 15 years compared to only 69% for HR. This means that although it may be a little slower, more of our respondents believed Financial Services were likely to fall victim to AI in the future. This might be due to the intense mathematical analytics that are used in this sector which can potentially be more effectively calculated by AI, especially with the growth of Big Data. Whereas those in HR might still believe that despite the help AI can give for recruitment, humans still need to personally evaluate interactions, as we discussed in our article on Big Data decision making last year. [Read our article on 'big data decision making']. Therefore our results showed a slight split between those who believe the transition in HR and Retail will be very quick compared to those who fundamentally still see the sector as in need of human interaction that AI cannot properly provide.

Nevertheless, there is some hope for humans in the AI world in the consulting sector...!  32% of respondents said it would take over 20 years or not at all for AI to overtake us. This shows that this sector is still more of a people based industry, reliant on human assessments and knowledge rather than machine learning. We agreed with this assessment in our research article 'the new realm of Big Data analytics' - although different forms of AI are being used by consultants to improve their analysis, it is still up to the consultant themselves to interpret this data most effectively for their clients [Read our article on 'The new realm of Big Data analytics'].

Our survey seems to roughly correlate with a recent global survey in June this year by the Future of Humanity Institute at the University of Oxford. The experts believed that AI will outperform humans in translating languages by 2024, writing school essays by 2026, driving a lorry by 2027, working in retail by 2031, writing a best-selling book by 2049 and even working as a surgeon by 2053.  The transition for retail to be outperformed by AI was slightly slower than the majority of our respondents believed, but still within the next 15 years.

But despite this, the experts also argued that there is no need to worry that AI will actually take your job anytime soon. The estimate was at least 120 years for AI to really have a serious chance of removing the need for humans in various tasks. An 'intelligence explosion' was deemed highly improbable, though not impossible. However, Noel Sharkey, a robotics and AI expert at Sheffield University, argues that although it is in many ways inevitable that machines will be able to outperform humans in many tasks, the question still remained on whether the technology is really comparable to a human interaction, just as our results for the HR sector potentially proved. Even if it is possible for AI to be used to maximise efficiency, humans may still reject it in favour of humans making more meaningful decisions rather than simply based on algorithms. 

Although all this artificial intelligence talk may sound like the stuff of sci-fi movies, many sectors are already using it in their work and we encounter it as consumers too with customer service 'chatbots', Siri on our smartphones and the tailored adverts on our Facebook news feed from the digital footprint we create online. A Narrative Science survey found last year that 38% of enterprises are already using AI, growing to 62% by 2018. Forrester Research predicted a greater than 300% increase in investment in artificial intelligence in 2017 compared with 2016   and it has been announced that Chinese investors are now pouring money into AI-focused start-ups, the Chinese government pledging to invest about $15 billion next year in 2018.   In fact at the Google I/O ’17 conference the hosts even boasted that they have achieved 'AI inception' with their AutoML system that actually creates better AI systems itself and there was alarm in the media that an AI system at Facebook was creating its own language. 

However, although things such as tailored advertising are useful for marketing companies, sometimes it could have adverse effects. Most famously there has been a big divide between tech giants Elon Musk and Mark Zuckerberg on the positives versus the negatives of AI. John Bara, president and CMO of a leading predictive marketing technology company points out that tailored news recommendations and adverts used for 'paid per click' (PPC) advertising also means that we end up getting a less balanced point of view as the messages and images you encounter online are all ones you already agree with and follow your preferences. Some even have argued that this has contributed to the unexpected events of Brexit and Donald Trump's Presidential campaign success.  

Nevertheless, the Future of Humanity research concluded that the probability of a bad outcome for the human race from the advancement of AI was low at only 10%. Either way, survey results about the future are really speculation after we get out of the 5-10 year foreseeable future range – we do not know what advances might be yielded from the increased investment in this field or how long they will take. At the moment, it is still humans who are needed to create the new AI inventions. 

When you think that even now in 2017 we are starting to have AI presented to us in a growing variety of ways, it is easy to see how this will quickly increase as more and more companies try to make a lucrative breakthrough like the Wi-Fi, smartphone and social media revolution that has transformed people's working and personal lives in the last decade. After all, it was only in 2008 that HTC launched the first smartphone on the Android operating system that now powers 352 million phones. So what can we expect in another 8 years?

Overall, 2032 looks like a big milestone for us to look to in terms of AI capability. It will certainly be interesting to see how this all pans out – perhaps it will be harder than expected for AI to gain exponential working knowledge... or maybe our estimates are too reserved and it will be even quicker.

Read the more from the Question of the Month series HERE

Have your say in our latest Question of the Month here!