I’ve had several pieces of feedback from readers of last week’s blog, which has led me to add an extra blog into this short series about where AI could be taking us.
Someone else mentioned the spoof nonsense about global warming. Yes, well, if you are watching the video version of this blog you will see I am wearing a pullover. We are in the middle of May in the Algarve. I have lived here for twenty-five years and this is the first time I have had to wear all this gear in May. Usually summer starts in May and we go into shorts and loose shirts, and very little else. I am now wearing a tee-shirt, shirt and pullover. And we are suffering from global warming? Big joke. North America has apparently had its coldest winter in living memory. We had a similar situation here in Europe a couple of years back. And as far as the statistics go Europe is still not quite as warm as it was in Chaucer’s time. On top of all that, more government statistics tell us the ice depth in the Arctic is the deepest in living memory.
Politicians think we are all mugs. To quote something from the rock opera Tommy “We’re not gonna take it”.
After writing last week’s blog about the way AI is going to put rather a lot of people out of work, a helpful reader referred me to a blog entry by Marc Andreesen.
I first came across Marc rather a long time ago when I was briefly employed writing html code. I had been having fun fronting a pop band for the previous decade, but had suffered an arm injury which put my live performances at serious risk so I was looking for an alternative way to make money and have fun. I thought writing code would open up rather a lot of intriguing opportunities, and one day I will tell the story of my first attempt to create a board game. I’m sure all the guys who now successfully make money in the gaming industry started off by making ludicrous mistakes, just as I did, but those mistakes do also give others a lot of entertainment.
The year was probably somewhere around 1991 or 1992. Marc was experimenting with html, and trying to come up with a better browser. It’s so long ago that I’ve forgotten the boring names of 1980s browsers, but they were all text based. That didn’t suit Marc, and he sent round a suggestion to the various coding communities suggesting an addition to the codex to allow the use of images. That led to the first browser that could display visuals. That was quickly followed by the game changer browser, Netscape Navigator. And, as they say, the rest is history. I think he now runs a very successful venture capital firm, so he is clearly a person to be taken seriously.
So much for the introduction, what about the content?
Well, he wrote a blog claiming that AI was not about to put people out of work. Over the years one of us will be proved wrong. That’s ok, but I shall call him out on this one.
First, he bleats on about Luddites. I am not a Luddite. I think AI is very interesting, and I’ve written a book in my Join the Girls series of novels which has as a main character, an AI home help. Unfortunately she comes to a sticky end (or do I mean she has an accident and has to go in for repairs?). My AI was introduced into a sort of normal household but she was so different under the hood from the rest of the household that pre-conceptions about other people got in the way. If you want to see what I mean, the book is available from Amazon. I’ve even reduced the price to $0.99 for a few days, so it’s an easy buy. Here are the links:
UK: https://www.amazon.co.uk/dp/B0CTHQ83H5
US: https://www.amazon.com/dp/B0CTHQ83H5
All good inventions add to our stock of goodies, and AI can be added to that class. There are usually downsides to inventions as well, and no doubt AI will have its fair share, but it is definitely going to put people out of work, especially when linked to blockchain technology. I will elaborate on that point later, but for this blog I wish to take friendly issue with Marc.
An AI will do what it is programmed to do. Program it to kill people, then that is what it will do. Program it badly and it will screw up. There is the old adage: Garbage in leads to garbage out.
Marc quickly moves on to give a few examples of what benefits there will be from the development of AI. But unfortunately, almost immediately falls into the trap of thinking about the future by imposing old ideas upon that future rather than using the characteristics of the new technology to envision a new future.
He says, “Every child will have an AI tutor”. Somehow I doubt it. Some children may have an AI that is a go-to for help in coping with things, much as every child may already have books to go to for help. That does not mean the child will use that help, or interpret it any better than interpreting books, or any other form of education that may be put before them. And there is the possibility that rather a lot of AIs will be used to purvey misinformation.
I maintain there will be no change on that front, just a proliferation of existing methodologies.
He also claims, “Every scientist will have an AI assistant/collaborator/partner that will greatly expand their scope of scientific research and achievement.” Granted. What now takes scientists, and others, a lifetime to gather information, test it, and come to conclusions, can be done in days or minutes by an AI. That’s good news.
But hold on. That puts a lot of researchers out of business. After all, the AI can do the legwork. If an AI can do the first researcher’s work much quicker and better, then a second AI can do a second researcher’s work quicker and better. Who needs all the extra researchers? In other words, it is likely human researchers will no longer be working in that lab.
Here we immediately come upon a serious problem. Marc, IMHO, does what so many people do in life. They add in all the plusses, and forget to subtract the minuses.
An AI is a machine that can function almost without stopping. Obviously it depends on the type of AI, but there is no reason why an AI cant function almost non-stop. In other words in an average week containing 168 hours the average AI should be able to work at least 160 hours whereas a human doing a traditional job requires meal breaks, sleep, recreation, visits to the loo, time off for illness, and so on, including good old fashioned skiving, and doesn’t work at all at weekends. That human also requires payment, and is likely to be far less efficient and careful than an AI, and only puts in about 30 hours work in the same time, and a human brain works considerably slower than an AI. In short an AI probably does fifty times the work of an average human. Why would anyone want to employ a human?
Why would an employer want to employ 50+ less efficient and more costly workers rather than one more reliable, better endowed AI?
If I put on my employer’s hat, I’m going to buy the machine and sack the humans.
In previous employment shifts we have seen alternative jobs appearing to cope with the requirements of changes in the work force. If someone invents a car, that puts the farrier out of business, but it means someone has to repair the car when it goes wrong, and someone else needs to repair it after an accident, and so on. More inventions lead to improvements, and to changes, and in the past those changes have led to an expansion of work opportunities.
The trouble with that scenario is that AIs can repair themselves, and probably produce more of the same. No-one is claiming there is likely to be 100% human unemployment, but it is going to be significant. As we go through this topic I will put some figures on the possible losses.
I mentioned blockchains earlier on. If we add blockchains into the mix, the outlook for human work opportunities looks grim indeed. But I’ll talk about that in a later blog.
Anybody reading Marc’s article will realise that I have barely started discussing this matter. I hope you’ll stay with me as I delve further into this subject. This is indeed a subject which is going to be crucial to anyone more than a decade away from retirement age.
Show Notes:
The AI Girl: UK: https://www.amazon.co.uk/dp/B0CTHQ83H5
US: https://www.amazon.com/dp/B0CTHQ83H5
The original blog by Marc Andreesen:
Share this post