To be clear, I personally lean towards being an artificial intelligence (AI) optimist, and frankly I’m incredibly excited about the potential for AI to transform our society. But I also recognize that there are potential downsides to AI as well, and one of those downsides is that over time some people, perhaps many people, will lose their jobs due to automation. When discussing this issue with numerous people I have noticed that some are in denial of the risk that AI or robotics poses to their livelihoods. In all cases they have made at least one, and usually several, of the following mistakes in their thinking.
Mistake #1: GPT 3.5 wasn’t perfect, so I don’t have to worry
This is absolutely true. But, the fact is that GPT3.5 was still incredible technology that provided valuable functionality. When Chat-GPT was released into the marketplace it garnered an incredible amount of attention, and rightfully so. But, because many people have unrealistic expectations of AI, expectations that weren’t met by the AI-based solutions made available to them, these people are now underestimating the potential of AI because the first release that they saw wasn’t perfect. Given that GPT 4 is even better than GPT 3.5, arguably an order of magnitude better, these people now run the risk of being blindsided by this improved technology. And it will only get better over time.
Mistake #2: GPT can’t do X (or at least not well), so I don’t have to worry
Yes, there are many things that AI can’t yet do, or at least can’t yet do well. But remember the following:
- GPT isn’t the only game in town. The AI sector is a highly competitive space, and as a result there is a myriad of AI-based technologies to consider. So if GPT can’t do what you want, maybe something else can.
- Prompting is a skill. If you’re like me, you may not be that good at prompting. In many cases this issue isn’t that the AI platform can’t do what you want, instead you don’t know how to ask it for what you want.
- GPT is quickly evolving. As we have seen with the recent release of GPT 4, there have been major improvements compared to the previous release. Expect this to continue over time.
Mistake #3: The unemployment rate is low, so I don’t have to worry
This is absolutely true, but as history has shown this sort of thing never lasts for long. At least in North America, the current shortage of people is due to the baby boomer generation retiring, some of the older Gen Xers taking early retirement, and sadly losing a swathe of people due to COVID-19. Bottom line is that I wouldn’t want to count on this bonanza of employment opportunities lasting too long.
The difficulty to find qualified people increases the motivation for organizations to automate their processes even further to address this shortfall. Instead of filling positions with people, organizations will automate/augment portions of their processes so as to free up people to do they work that they haven’t yet automated. This reconfiguration effectively reduces the number of unfilled positions. In short, many of the job openings that exist today will disappear tomorrow.
Mistake #4: There are aspects of what I do that can’t easily be automated by AI, so I don’t have to worry
This is true for virtually every job type. But there are many aspects of what you do that can be automated, and many of them are sufficiently common which implies there is a market for an automated solution. I believe that few jobs will be automated away all at once. I am absolutely certain that portions of jobs will be automated away over a relatively short period of time (think by the end of this decade), that organizations will reconfigure how they work so that they don’t need as many people as they have today. In short, many jobs will disappear through “death by 1,000 cuts” rather than “death by terminator.” More on this in a future article.
Mistake #5: AI will augment what I do, so I don’t have to worry
The proponents of AI are very clear about the potential of AI to augment people’s work, and there is no doubt that this is happening in many organizations in a very wide range of domains. Work augmentation is a very good thing, but will you be given the opportunity to do so? Don’t assume that everyone is going to invited into the lifeboat. We’ve all seen organizations management organized people into those who would continue with the organization after the new tech was adopted and those who would be made redundant. Why will it be different with AI-based automation? Furthermore, are you stepping up to learn about AI, are you identifying ways to apply AI to augment your work, and are you involved with bringing in this new tooling? Doing these things will help to increase the chance that you’re included in the group of people who are kept on in roles that are augmented by AI.
For some interesting insights into the potential for AI to affect jobs, see the research study GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models.
Mistake #6: AI will create new jobs we can’t yet imagine, so I don’t have to worry
This is absolutely true, for example “prompt engineer” likely wasn’t on anyone’s radar at the start of 2022. New types of jobs will surely be created over time. But, there are some very important questions that tend to get glossed over:
- Will there be as many “new jobs” created by AI as are destroyed by AI?
- Will these new jobs be created at about the same time, or better yet before, the old jobs are destroyed?
- Will you be qualified for any of these new jobs, or at least able to become qualified for them?
- Will you be interested in these new jobs, e.g. do you really want to become a prompt engineer?
- Will you be hired into one of these new jobs, even if you are qualified and interested?
- How long will these new jobs last? For example, I suspect prompt engineering will have an 18-to-24 month window at best.
Mistake #7: I can always find another job, so I don’t have to worry
Don’t count on this. Granted, the job market is great right now (see my earlier point). But, as Calum Chace describes in The Economic Singularity, we’re getting to the point where AI and robotics is being applied in across all industries to automate many things. As a result we are likely to see a relatively short period of time, say 3-5 years, where jobs are whittled away across all industries simultaneously. Instead of one or two industries seeing a job downturn, it will happen across most industries at the same time. In the past you could lose your white collar job in industry X, but find a similar position in industry Y after a bit of effort. With the AI revolution, we may discover that we lose our job in industry X and find ourselves competing for a dwindling number of jobs with people who have similarly lost their jobs in industry A, industry B, and so on.
Concluding Thoughts
We need to stop fooling ourselves regarding the threat of AI to our employment opportunities. It is human nature to play down threats, I get it. Bad things only happen to other people, until they happen to us too.
Yes, the 2020s are going to prove to be an exciting decade to live in. This may not be as good as it sounds. I hope this article has given you food for thought. Let’s be careful out there. 😉