AI: What Could Possibly Go Wrong?

As a teenager in the 80s I wrote an essay on how robots would end up being the next stage in human evolution.

I had grown up reading great novels by people like Isaac Asimov and Arthur C. Clark, my family were all engineers and I was studying science and technology at college. The future seemed very exciting (which indeed it turned out to be) with huge possibilities for human development.

But I got one crucial thing wrong, let me explain.

The vision I had, based on ideas from many other people far more clever than I, involved machines that extended human ability. This included powered exoskeletons to improve strength and stamina as well as increasing or decreasing the scale of movement as appropriate, increasing dexterity and having specialist tools fitted to do those tricky jobs that only a superhero could do.

But it also involved increasing brain power by having extra memory and computing ability to extend our brain’s capability far beyond natural limits. And that, to an extent, is already happening by way of the smart phone. I have instant access to the world of information, I have maps so I know my way round any town, I have links to thousands of people I’ve never met, I can talk to people all over the world whilst I’m walking down the street etc. We may be used to it, but actually it’s pretty impressive.

As a kid my vision went further than this, to a point where I had an extension to my brain built in, with the robot limbs attached to my body so I became one with the technology that made me stronger, faster, smarter etc. I figured everyone would have this eventually and this would be the next phase of human development. And it almost was.

Let me give you one, very specific, example. CAD, computer aided design, has made designing and making things so much easier and improved quality too. Early systems I used in the ’90s were basic tools that replaced pencil and paper drawings, this was great, then they got steadily better and added functions. Along came analysis systems such as FEA that can not only draw your design out but actually simulate forces going through it and identify where weak point may be, wow, this saved loads of time in testing and made is much easier to design light and strong components. Previously a design engineer would use experience and basic design principals to draw something up, then it would be tested and any failures analysed in order to improve the design. So now the skill and experience was in the machine, allowing CAD users with little mechanical knowledge to design fairly good components. This improved quickly and now these design packages can actually take a vague concept and do all the design work themselves, so it takes far fewer engineers to get a new thing designed and built.

So what happened was I started out with a simple tool that helped me draw, then it improved my design, but now it can do all the design and the machine no longer needs me to be there.

This is happening with AI driven Expert Systems, which pick up the knowledge and experience of many experts and synthesise it into very powerful knowledge systems that can learn from their own mistakes. These are better than any one single human expert. They are replacing Pilots, doctors, teachers, designers, engineers and are also replacing artists. Yes, an expert system can be set up to write new music, paint pictures and write stories to a very acceptable level, and they are getting better all the time.

By replacing humans in a company costs can be dramatically lowered, 24 hour running is possible, there are no strikes or HR problems, you don’t need buildings with heating or air con to the same extent. The financial pressure to implement these systems is huge. And this is driving investment into AI and causing it to be implemented without mitigation of the adverse effects on the people who no longer have jobs.

So whilst many people foresaw that machines would bring greater powers to us, what I missed was that once they got good enough they wouldn’t need me. The human element becomes redundant.

Now, what happens when there are very few jobs available? Mass unemployment is already creeping into the western world, and what the politicians don’t seem to be telling us is that this is because there are less jobs even though there are more companies who are doing more business than ever before.

Manual labour replaced with machines (just look at farming, even the combine harvesters are robots now), knowledge and skills replaced by AI (how long before expert systems replace judges in our courts?). Where do we fit in? Where does my young son fit in when he grows up in this world?

There are other problems too. There is also the issue of corruption. Computer systems get hacked, there are bugs and viruses, so total reliance on these systems is very dangerous. But to have a human back up needs the investment in people, training, facilities etc. that AI has just made redundant.

Then there is the whole rotten cesspit of autonomous military systems. Drones that decide who to kill, tanks without crews, smart missiles. This is stuff that already exists and is getting more sophisticated all the time, and most of the cutting edge stuff is obviously developed in secret.

But also there is the interesting aspect of group intelligence, because the internet is connected with millions of machines, smart systems can be spread across many physical platforms, the Cloud as it has become known. So we have a multitude of smart systems that have potential access to all the online knowledge, plus bank accounts, medical records, criminal records, documentation showing who owns your house and your car, who the legal parents of your child are, your nationality, passport, your social media, your pictures etc. A malicious system could hack your entire life, set up a criminal record and get you locked up. The net also has access to the physical world thanks to the Internet of Things, such as nuclear power stations, flood defences, gas supply and even where that robot combine harvester goes. A hack to Google maps might send thousands of motorists into one city centre location to cause gridlock, or to confound the response to a terrorist attack.

There is absolutely no control over any of this.

Our society is based on a magic thing called freedom, trying to precisely define it is impossible and probably pointless, but we all have a vague idea it means we can choose our own path in life as long as we don’t do very bad things. We choose what to study, or if to study. We choose what to work in, or indeed not to work in. We choose our partners, where we live (although that’s often dictated by where to work), what to eat etc.

This means that government has a largely reactive way of managing problems, western governments don’t like to get too involved with running things. This means that companies have a large amount of freedom to develop what ever they want, which has generally been a good thing. But this is different, this is one of those things that is about the very future of our species.

We need a plan, we need to agree what direction society goes, how it uses technology to benefit us all. We need control over this situation before something ‘very bad’ happens.

Anyway, that’s my opinion. Hope I’m wrong. But this bloke seems to have the same idea.

Scroll to Top