Discussion Forums » In the News

Artificial Intelligence Article in Times

    • 82 posts
    January 3, 2018 10:14 AM GMT

    The following article appeared in the Times on New Year's day 

    https://www.thetimes.co.uk/edition/comment/no-need-to-fear-the-rise-of-the-machines-hkpz5p7ld style="color: #b00000;"> sadly I think it is erroneous and going to be typical of the debate we get about AI as it becomes more and more a mainstream topic below I've taken a close look at it:

    No need to fear the rise of the machines - Sunday Times Article on AI by Matt Ridley

    Matt Ridley’s article is a hopeful one that AI will continue to augment humans rather than replace them and we could all enjoy the benefits.

    He uses the argument that in the 1960s there was a disagreement in MIT about what computers would achieve and says that one faction championed AI, believing that computers would gradually replace human beings. The other championed human-computer symbiosis believing that computers would augment human beings. Matt Ridley suggests that looking around us the symbiotic argument has won.

    Matt also repeatedly uses the argument that our experience of the past with new technologies is a good guide to the future.

    So I believe his article is largely based on two fallacies.

    Firstly the symbiotic argument hasn’t won because the game is only just starting. AI is only just being deployed and in fairly simple form. It’s certainly true that in the 60s and 70s most of us thought AI would advance very quickly but it has taken a generation to get there.

    Secondly the past cannot always be relied on as a good guide to the future. Certainly it raises the question that has to be answered, why won’t it be the same this time? The answer is that AI is fundamentally different to computers to date. Those computers and the internet have given us an infrastructure for rapidly deploying AI and AI inherently will be able to replace any job, even the highest value jobs, at some point in time. Importantly it is also inherently capable of replacing any new jobs created. The replacement is unlikely to be complete replacement of the existing way of doing things but replacement by new and better ways of doing things.

    Certainly I would agree that augmentation of jobs is likely to precede full replacement of jobs in many cases. However it is all a question of timeline. Ultimately the potential is there if we choose to allow it. If we do nothing economics will dictate what happens. If you are starting a new business, why would you use labour rather than AI if you don’t need to? You will be able to do whatever it is with both higher quality and lower cost and to undercut any competitors who are still using labour. Even in person to person jobs such as sales, would we prefer to be sold to by a salesman or use a system that took our requirements and matched them to the best system on the market using Meerkat AI.

    Interestingly full symbiosis with humans through brain implants etc could be a direction to go, but that’s another whole story.

    A lot of what is said in the article is likely to be applicable in the early days of AI but I think it fails to appreciate the potential of AI e.g. suggesting it is unlikely to impact services, or that we won’t have totally autonomous cars. The article relies on previous predictions about replacing human jobs by technology being wrong – this has already been dealt with to some extent but I think there is a bit of ‘the boy who cried wolf’ syndrome involved.

    Where I think the article does raise important issues is around jobs. It suggests that middle skill jobs will be impacted first and that based on experience over the last 20 years that both lower skill jobs and higher skill jobs will result. I think this is quite likely in the early days but it will be economics and ease of deployment that determine which jobs go. Certainly middle ranking jobs can be expensive but low skill jobs where large numbers of people are involved are also likely to go.

    The article rightly raises the point that capital based economic change will make society richer in aggregate but that this will create a problem of distribution i.e. those who can provide labour but don’t own capital might have inadequate means of making a reasonable living.

    One could take a lot of what I’ve said to date as leading to a doomsday or social breakdown scenario but actually I think there is a huge message of hope as long as we recognise potential issues early and plan for them. In my view AI holds out the amazing possibility of solving most of our big issues, from the NHS to Care Homes and Housing, tackling both quality and cost at the same time.

    What we must do so we don’t go down the dystopian path, is start planning now. Looking at both some of the obvious consequences but also searching for possible unintended consequences and addressing both, ahead of the game.  It doesn’t matter whether all this happens in the next 20 years or the next 50 but it will happen.

    I was brought up with the idea that the advent of computers would allow us all, by now, to be working shorter hours for more money. Instead most of the benefit has gone to owners of the technology. This time around we need to plan how we all benefit and how we manage the costs of the disruption it will cause. That involves looking at the tax system for AI deployment both via the internet and in physical form when coupled with robotics, so that all the benefit doesn’t end up in a few totally dominant international corporations.

    Do nothing and we will have a dystopian future, plan ahead and act early and we can have a utopian future!