Leaders at the beck and call of ChatGPT ? (3/4)

How is AI impacting talent acquisition, management, and retention ?

In his 2016 book, AI Superpowers, Kai-Fu Lee identified three buckets of solutions to AI induced job losses. They were the three R’s of retraining workers, reducing work hours, and redistributing income. The first solution is premised on the idea that AI will shift what skills are in demand over time; the second believed that there will be a permanent reduction in the demand for human labor and that this can be absorbed by moving to a three or four day week; and the last is underpinned by a conviction that AI will displace workers in such a way that no amount of training or tweaking of hours will be sufficient.

The implicit assumption behind this third question from our ChatGPT prompt, is that the solution to the societal dislocation caused by AI is in the first of these three camps, “retraining workers”. Either you find new workers with more up to date skills, or companies should re-skill their existing workers.

However, I believe that this question is missing a bigger issue. Aside from the fact that its proposed solution set is too narrow, it also implicitly assumes that generative AI is simply a tool that people can be trained in. This echoes people such as Sam Altman who have been quoted saying things such as, “it’s not a dangerous AI, it’s just a tool. It’s a blinking cursor. It’s waiting for you to put in what you want it to do”.

However, I would contend that ChatGPT is not just a tool. People have already figured out how to take OpenAI’s GPT-4 and make it run itself in a loop. You can give it a goal such as, “make as much money as possible”, and then it starts to figure out its own plan and execute that plan by calling out to other services such as TaskRabbit or Craigslist after having been connected by its creator to a bank account. People have taken that blinking cursor and turned it into an autonomous loop or agent that takes actions in the real world.

The fact that such agents start executing on objectives given to them by their creators without any hesitation, challenge or chiding is concerning. This is even more the case when you consider how easy it is for this technology to proliferate, and how the associated risks increase if those agents start to leverage the un-sanitized versions of large language models that are becoming more and more common and accessible on coding websites such as GitHub.

The key takeaway: Many of these ChatGPT powered autonomous agents, if not already doing so, will be capable of doing things that will probably require some form of license in the future. Responsible business leaders should indeed think about how they train and modernize their workforce, but they should also be actively contributing to global, regional, and local discussions about regulating AI and ensuring that a new level playing field is created for all companies and workers across the globe.

Executives just focused on how best to reskill existing workers or learning how to hire next generation skills, will likely miss the more important discussions impacting talent such as changes in working practices, income redistribution and helping people find a new purpose in life. If they don’t play an active role in these bigger conversations, then they risk becoming increasingly irrelevant.