Globalisation Replayed: Will AI Leave Workers Behind?

Do you remember the postcards with the quote, ‘To err is human but to really foul things up you need a computer'? As government ministers bounce around like excitable spaniels, panting ‘AI good’, I’m sure I’m not the only one hearing alarm bells.

To make it clear, I’m no Luddite: when it’s doing things that humans can’t – like voice recreation for people with MND – AI is welcome and astounding. But it’s not just being used to enhance human ability, it’s increasingly being used as a replacement for human activity; hence the alarm bells.

I don’t think I’m being unreasonable in thinking there should be two conditions when it comes to using AI: first, government and industry must not forget human beings and second, the data it’s trained on must be robust. 

Already, there is a sense that the wheels are coming off when it comes to the data. I admit to some mild irritation when it got some niche medieval genealogy wrong – whoever had inputted the data had confused a son for a brother. A minor mistake but telling: without a human paying attention to detail, AI can’t get things right. Now, there’s a recipe for chaos… particularly when someone decides to train AI on tweets.

Maybe the word ‘intelligence’ is the issue. It confuses people into believing that AI can think. AI cannot think; it can only follow instructions. It can condense shedloads of data into a neat summary but, if that data is in any way, shape or form wrong, AI will either incorporate the mistake (with bells on) or reject it. Unlike a human, it won’t just recognise and ignore a mistake unless it has been instructed to.

In an ideal world, rather than expecting applicants to demonstrate the psychic skills that would enable them to pick out the winning lottery numbers, recruiters would provide multiple tags for AI to screen against. They would also create decision logs, so if they’re not getting the results they expect they can see why.

Surely that would make more sense than posting a job advert, including an obsolete qualification, and instructing AI to reject any candidates who don’t include it in their application? And before you think there’s no way that would happen, it did – at a tech firm in India.

Training AI on CVs submitted by successful candidates over the years sounds like a great idea...until it automatically incorporated historical hiring biases. Amazon scrapped that hiring tool before it had left the test phase but how much of a cautionary tale it has proved is debatable.

In fact, caution appears to have been thrown to the winds when it comes to AI. We have adverts with AI models, governments agreeing that copyright should be ignored for the purposes of training AI (as a former Director of Public Prosecutions, you would have thought the current Prime Minister would have a passing understanding of the concept of theft, but apparently not) and AI gobbling up entry-level and creative jobs.

Meanwhile, students are using AI to write their essays for them, and both applicants and employers are using it to aid recruitment, meaning that AI is screening AI and humans are bypassed completely. A reminder, if one were ever needed, that just because you can do something, doesn’t mean you should.

And here’s where the government’s ‘AI good’ policy is more alarm siren than bell.

Remember when government ministers told us globalisation was good? To keep the cost of goods down, manufacturing would take place in countries where labour was cheaper. Loads of new jobs would be created in the UK, which the people displaced by having their jobs shipped abroad could do. Except that those new jobs either didn’t appear in the places that had been deindustrialised or couldn’t be done by the people who were displaced because their skills lay in making things.

I spent the early part of my career on newspapers in the north of England. I’ve seen what happens when people in government start behaving like excitable spaniels. When jobs aren’t replaced, there’s no money in the areas they’ve disappeared from. Our economy is based on growth – the idea being that everyone buys stuff, which keeps everyone else going. (I’m ignoring the difficulty of infinite growth on a planet with finite resources for the purposes of this post.) But if you have no money, you can’t be a consumer. Those who can, move in search of better opportunities, leaving behind those who can’t. Eventually all hope and most opportunity is gone and you’re branded ‘work-shy’ by people who simply had the luck to be born in a different part of the country.

‘Globalisation good’ didn’t have a precedent within living memory. We now know that, as well as the strategy, there needs to be a plan, and that leaving it to the markets to decide doesn’t hack it. 

In the current rush to embrace AI for absolutely everything, we can see companies getting rid of entry level jobs. It doesn’t take a genius to see that, in 15 years’ time, they’ll all be complaining that they can’t find anyone with the necessary experience to be a manager. That’s what lack of planning and short-term thinking does for you.

What’s the point of an ‘AI good’ strategy if you don’t have a plan – or at the very least a framework – to make it work? I want to see that the current government is doing its thinking when it comes to working out the jobs that will go and those that must stay, what the new jobs are, and where in the country they will be. I want to see that they’ve adapted the curriculum and are working with further and higher education so young people have the necessary skills. I want to see that they’re supporting businesses with training courses to upskill existing workers.

I especially want the government to take some responsibility for ensuring the over-48s have a place in the ‘AI good’ world. After all, with the raising of the state retirement age and the falling birth rate, we’ll be the largest section of the workforce by 2030. Forgetting about us won’t come cheap in any sense of the word.

This is tough stuff – the pace at which AI is moving means that skills needed today will almost certainly be obsolete in three years’ time. Good luck with that trajectory when planning a degree course.

One area I’m willing to bet will see substantial growth is the field of AI law. That’s not just the fields of copyright, data protection and contract, but also new areas such as the consequences of advertising a face cream on an AI model, which doesn’t have skin. Should that be legal? Think of all the billable hours in working that one out.

The government is right that we cannot ignore AI and that there are elements of it that should be embraced, but we’ve seen what happened with globalisation. The governments of the time could be called naive. When it comes to AI, the current government, and those of the future, have no such excuse: if they don’t act, they’ll be guilty of negligence.

Coming soon – we’ll have more on job seeking in an AI-world and what we’d like to see in the future.