By Scott Hamilton
I recently read an article online about the Godfather of Artificial Intelligence and his take on the future. The renowned scientist and AI pioneer Geoffery Hinton spoke at “Collision” in Toronto, Canada, where he warned big tech companies to look out for, and work to prevent, an AI doomsday. In his talk, Hinton outlines several areas where AI is already impacting human life and gave fairly strong warnings that there may come a time that AI will attempt to eliminate mankind.
A lot of Hinton’s speaking points I have written about before, including the fact that a properly written AI has the potential to take on the attributes of a living being. We can already see with advanced algorithms like ChatGTP that AIs are very capable at forming seemingly coherent and independent thought. Just ask ChatGTP a few questions and you can already feel like you are talking to a human. ChatGTP even makes mistakes like a human might make, including mathematical errors.
I would say the thing that struck me the most from Hinton’s speech was when he talked about the jobs that AI would take away from humans. We already see AI being used to write news stories, computer code, songs and movie scripts. There are chains of fast food restaurants working to replace the back line with robotic workers. A majority of manufacturing facilities today run nearly entirely on robotics, and nearly every movie in the last 20 years has had computer generated special effects and even actors. That leads one to think, what kind of job is safe from being taken over by robotics and AI?
According to Hinton, there is really only one job he considers to be future proof. The job of a plumber. So if you want to remain employed, as robots and AI take over more and more roles, including scientific research and computer engineering, you might want to consider a trade. Hinton mentions that custom carpentry is also a fairly safe choice, but points out that it is quite easy for a robot to build furniture, cabinets, and even entire houses. What becomes difficult for AIs are tasks that require thinking outside the box.
Examples of such jobs are building custom shelves in older homes with uneven walls, repairing existing structures and furniture, and plumbing, the most difficult. Next time you are in a hardware store, go take a quick look at the plumbing section. You will find that there are nearly always multiple aisles of plumbing fittings, fixtures and couplers. A quick check of Menards’ website shows a selection of twelve different kinds of pipe and fourteen different types of fittings. These pipes and fittings come in more than twenty sizes and each size of pipe can have multiple types of connections. I cannot even count the total number of individual fittings and adaptors. This is not a problem for a new plumbing installation, but if you go out on a repair job, you can get into a mess.
Any house more than five-years-old has the potential of having had multiple plumbing repairs and most plumbers will attempt to fix the issues with the supplies on the truck; as a result there can be a mix of piping types, sizes and threads on any job site. This requires quick adaptability, which is something that current AIs are not very good at accomplishing, and not likely to be good at in the near future.
The biggest take away and worry from Hinton’s talk was his quote, “Right now there’s 99 very smart people trying to make AI better and one very smart person trying to figure out how to stop it from taking over.”
While I am sure it is an exaggeration, it is not far from the truth.
“Before it’s smarter than us, I think the people developing it [AI] should be encouraged to put a lot of work into understanding how it might go wrong… and I think the government could maybe encourage the big companies developing it to put considerable resources [into this],” said Hinton.
I agree with his premise entirely. Until next week, stay safe and learn something new.
Scott Hamilton is an Expert in Emerging Technologies at ATOS and can be reached with questions and comments via email to firstname.lastname@example.org or through his website at https://www.techshepherd.org.