An AI's forecast for areas it will threaten
December 3, 2019 at 9:34 am #38671ParticipantEd P@edps
No I don’t mean crazy killer autonomous robots, though that is of course a worry. Rather, how should we counsel our kids/grand children in their paths through life. What paths/areas should we encourage and which ones should we flag with ‘Here be dragons’?
A daunting task, so it is useful that the Brookings Institute have produced this forecast using AI tools to make their predictions.
Its a long report and goes through both a sectorial and geographic analysis of every US job function.
Doing a very sweeping paraphrase of this noteworthy report, I came away with the impression that both tall poppies and lowly labourers were those most at risk.
AI enabled automation will result in general purpose machines capable of taking over generic repetitive jobs. ‘Pure’ task focussed (more expensive) AI, on the other hand, will be aimed at the better-paid, white-collar occupations as their elimination will be more easily cost justified.
If you want to encourage anything, I get the impression that ‘Learn a Trade’, ‘Creative arts’,or become a professional footballer are the things to do.
I’m not sure I agree with some of the weights that have been applied in the AI analysis. For example, Legal is shown as very low risk, while I would have thought that the legal area is over-ripe for full AI driven automation. Having found one weighting to dispute, there may well be more. Unfortunately AI is a prime example of “crap in = loads of BS out”. As a result rather than just dumb acceptance of the report please use it to stimulate your own analysis.
Criticism aside, if you are interested in a forecast of the future this is a report to read.
December 7, 2019 at 4:51 pm #38725
Via Bloomberg I picked up an article on the multi-million pound London based XTX Markets. This is a Trading Company/Market Maker that runs with ZERO Merchant Wankers, and is essentially staffed by a small team of mathematicians and computer programming/AI wonks. They probably have the highest productivity/profitability per head in the City and point to the future of City jobs that AI will gobble up.
Taking it one step further, other than a tiny London marketing office, there really is no need for such a team to be based in London. It could be anywhere nice with a friendly tax regime and a huge Internet pipe.December 7, 2019 at 5:29 pm #38727ParticipantBob Williams@bullstuff2Forumite Points: 13,458
The three major fields that I can see AI struggling with, at least for the next 5 to 10 years:
Prediction – what to do about futures, how to forecast trends with any expectation of accuracy.
Decision – how to instil/programme decision making, again with an expectation of accuracy. Some of the activities/industries that AI will have to work with, require decisions that will have repercussions for the Earth and all of humanity.
Control – how much control over processes and procedures, to hand over to any AI, in order for the results to be free of errors that may also have repercussions.
But the most difficult quality to emulate IMO, is how to instil the one characteristic that certain humans possess, which has always brought progress. (beneficial or otherwise). Intuition.
Positive results will also be affected by the humans and organisations who produce, manufacture and market AI devices and machines. As matters stand now, only multinational companies have the clout and the finance to do all that. They are not noted for their altruism. Some form of UN oversight and international cooperation will be necessary, I believe.
Good luck with that.
When the Thought Police arrive at your door, think -
I'm out.December 7, 2019 at 11:30 pm #38731
Actually Bob, one of the hardest challenges is one that HMG have just mandated must be done by any Financial/Government AI and that is proving and explaining the logic used in arriving at an AI’s decision. (very hard when ‘hidden’ factors are used in neural networks). Even then it is difficult to rationalise, for example many of the vaunted object identification programs do not actually see objects but instead find differences in textures – a recent discovery that explains many of the idiot outcomes from such programs.December 8, 2019 at 12:42 pm #38736Participantkeith with the teef@thinktankForumite Points: 2,727
I dont have much time for the term AI or AI.December 8, 2019 at 10:05 pm #38754ParticipantBob Williams@bullstuff2Forumite Points: 13,458
I dont have much time for the term AI or AI.
How do you refer to it Keith? It’s on the way, we have to call it something or it will become terribly confused, get angry and start turning off everything from satellites to traffic lights and your fridge freezer.😆 Civilisation will collapse and it will be all your fault!🙄😣😃
Ed, I would be happy with any AI that could find a way for the NHS Digital systems to talk to each other. They have only been trying since 2013, after all. Perhaps the fact that they launched for the first time on April 1st had something to do with it…
Just a small selection of the problems.
When the Thought Police arrive at your door, think -
I'm out.December 9, 2019 at 7:32 am #38757
Bob, AI will never sort out bad and lazy management, especially when they are given the authority to sabotage any progress that threatens their own fiefdoms. That can only be improved when someone is put in charge who has a clear vision of the requirements and has both power AND accountability over the final result.
Imo the NHS has been completely screwed over by the Conservative Government’s failures to grasp the nettle of Social Care. It is now too busy trying to drain a swamp of high priority issues and I’m afraid getting systems to interface is way down the list when they are still struggling with XP era computer kit and a dire shortage of staff caused by Brexit.December 9, 2019 at 4:59 pm #38763
- You must be logged in to reply to this topic.