A new study shows that imitative AI has not helped most workers become more productive. It has, in fact, burned them out and made them less productive. I am not surprised, but maybe not necessarily for the reasons long-term readers might think. This is only partly an imitative AI issue — it is mostly a business issue.
Make no mistake — this is at least partly an imitative AI issue. Imitative AI is not especially good at automation, either the kind that replaces people or the kind that augments people. It simply makes too many mistakes and is not consistent. Human beings are not good at coming in behind and double-checking automation systems. It is actually harder work to check the output of every action rather than creating the output yourself, especially when the issues are subtle.
The lack of consistency adds to the issue. Rember, imitative AI systems do not produce something because they understand the request. They merely calculate what they think should come next based on their training data as compared to the input. Those calculations can and do produce different output based on variances in prompts, timing, training data accessed, etc. So, they cannot be trusted to reproduce consistent output. If you cannot trust that something that needs to be the same every time is the same every time, then you aren’t really gaining much in the way of productivity by relying upon it.
The limitations of imitative AI should be clear to anyone with a passing understanding of how it works. However, the survey shows that business leaders apparently do not have such a clear understanding. Ninety-six percent of CEOs expect imitative AI to produce productivity gains, but almost half of employees say they have no ideas how they are supposed to use the tools to being with. The problem is, at least in part, indicative of how poorly modern businesses handle employees.
While imitative AI companies have consistently over-sold the abilities of their products, that is what salespeople have done since the first caveman tried to convince his cave mates that sharpened rocks would end bad hunts. It is the responsibility of the business leaders to wade through the nonsense and plan appropriately. In this case, the introduction of new tools obviously means training costs. People have to learn how to take advantage of the features of the products they are expected to use. That is basic common sense. But modern business is no longer good at common sense.
Business looks at employees as a cost, not an investment. Not all businesses, of course, but there is a distinct pressure to see the people who get work done as an impediment to profit rather than the source of profit. The idea that you invest in people is no longer common sense — it is more of a heresy. So, when companies that are already less inclined to help their workers becomes productive were presented with a tool that could do the work without training? It must have seemed like their kind of magic.
And magic is what they are seeking — the ability to produce profits without the need for any investment in human beings. The making of money without the need to have people run the business. But magic doesn’t exist in the real world. And wishes do not make reality bend. These businesses have failed to recognize why imitative AI likely cannot help them due to not understanding how their money is made. And those businesses that could find some benefits from imitative AI do not recognize the need to train their people on these tools. They look at the world and see an illusion. As a result, they fail to make correct choices, they burn their people out, and they lose the opportunity for whatever gains they could have made.
Imitative AI is not well suited to augmenting or replacing most jobs, that is true. But the larger issue hiding in this survey is just how poorly business leaders understand their own businesses and how little they understand what is required to actually be profitable. And no amount of fancy Clippy is going to solve either of those problems.
