PicoBlog

I’ve been doing a lot of pondering about how to think about generative AI. As a result, I’ve come up with a three paradoxes and two metaphors that might help you think more deeply about how it works, how to use it and its impact on the workplace.

One of the questions I always get about generative AI is, “won’t it leave a lot of people behind?” And the answer is…complicated.

For those that use it, generative AI is both a means of catching up and of getting ahead. That’s the paradox.

For example, a recent study on the use of a ChatGPT-based tool by 5,179 customer service agents showed a 14% increase in productivity. But more interesting than the productivity gain is that the biggest gains went not to the best agents but to the novices. Novice agents gained almost all the boost in performance while expert agents gained only minimally. This is what I mean by it being the great equalizer.

It can also help younger TV writers compete with those with more experience. In this Cognitive Revolution podcast, Hollywood writer Trey Kollmer talks about his use of ChatGPT to write scripts. In it, he speaks about how it is semi-useful for him. However, Trey goes on to say ChatGPT could be much more helpful for younger writers as they develop and create their scripts to ensure they have considered and addressed everything they need to for it to be complete before submitting it for consideration.

Lastly, in Higher Ed we discuss how college applicants can use ChatGPT to create better candidate essays. The essay is a means of getting a personal statement from the student and also test their ability to think logically and write persuasively. Thus, it is a problem if the applicants are having ChatGPT ghostwrite their essays. However, as another professor pointed out to me when I raised this issue, well-off candidates are already having their essays ghostwritten for them. ChatGPT just lets those less well-off to compete with wealthy applicants.

So in certain instances, generative AI can eliminate the gap between novices and experts and between the well-off and those less fortunate.

However, in other types of situations, this may not be the case. For example, if you are an expert in a more autonomous job (i.e., a knowledge worker), you can use generative AI to dramatically increase your productivity and creativity, leaving those who don’t use it far behind. You can do this by using generative AI to be your personal assistant and delegate tasks to it and to use it for brainstorming and other creative jobs.

The takeaways here are;

• If you are a novice, generative AI can help you compete with more experienced workers, however, if you’re a top performer in a non-autonomous job, you may now see the years of expertise devalued by generative AI.

• If you are an employer, you might be more willing to hire less experienced workers (and perhaps shed more experienced and costly ones) and trust AI to fill the gap.

• Also, from an employer standpoint, using generative AI in roles in which it improves novice performance means that your average results will improve and your performance distribution will tighten. That leads to better results with more consistency, which is very attractive.

• Back to employees, the big takeaway is that if you don’t want to be left behind it behooves you to up your generative AI skills. As the saying goes, “you won’t be replaced by AI. You will be replaced by someone who knows AI.”

We expect AI to be better than humans at traits such as accuracy and speed. However, in some roles, generative AI can be better at more human-like traits, such as empathy and patience.

One example; bedside manner. In this study, not only were chatbots answers to patient questions often superior to physician responses in quality, they were also better in empathy. This should not be totally surprising, as AI is infinitely patient and very polite, unlike humans who are on tight schedules.

Generative AI can also be extremely good at creative tasks, as I wrote in an earlier piece on Cross-Domain Thinking. Generative AI’s ability to brainstorm ideas and compare/contrast concepts at speed is impossible for humans to replicate. Instead, we need to take advantage of it.

In Hinduism an important god is Shiva. Shiva is viewed as the Destroyer but also as the Creator. He transforms the universe.

“Creative destruction” and transformation is something we also associate with capitalism, and generative AI plays a similar role.

Of course, generative AI creates content. It’s in the name - it generates things, as in “genesis.” It has also created new tools, use cases, jobs (“prompt engineer”) and firms. However, it will also destroy; websites that depended on Search will now be bypassed with answers direct from chatbots. Jobs, tools and companies will also go by the wayside as generative AI improves. Thus, generative AI, like Shiva, transforms; it transforms our world, our roles and ourselves. We are far from being able to see exactly how things will change but it’s clear many transformations will occur and be massive in impact.

We’ve all heard the metaphors of ChatGPT being “autocomplete on steroids” and a “stochastic parrot”. These metaphors do help many understand a bit about how ChatGPT and related types of generative AI work. However, I think those metaphors dramatically understate the capabilities of these tools, making them sound far more simplistic and less powerful than they are in reality.

There are two metaphors I find useful in explaining ChatGPT-like generative AI, the swiss army knife and the microwave.

Admittedly, this is not super-creative as a metaphor but it is still useful because ChatGPT et al can be used for so many different tasks. Those tasks range from the mundane (drafting an email) to sophisticated coding and cross-domain thinking.

Many people think of ChatGPT as something that replaces search but that is actually where it can hallucinate. Instead, one should use it more for providing creative ideas, learning how things work, outlining articles, etc.

Generative AI is fast. Very fast. And it radically changes how quickly you can accomplish formerly complex and time-consuming processes.

Consider the typical Problem-Solution process below. There are several steps in it and prior to the internet and search, the process was time and resource-intensive.

The internet and search shortened parts of the process; especially gathering data and identifying potential solutions. But much of the rest of the process was unchanged.

However, generative AI can be used in every step of the process, shrinking the amount of time needed to complete it by (a wild guesstimate) 80%. Just like a microwave, it cuts time drastically. Unlike the microwave, what comes out can be better than what is produced in the old way.

Hopefully, these paradoxes and metaphors have been helpful in thinking about generative AI. Perhaps they’ll also enable the explanation of the power of these tools more impactfully to those that aren’t familiar with them.

ncG1vNJzZmilkae4rq%2FNnqClpKljwLau0q2YnKNemLyue89oq6GdXam1s7HEZqeaqpGZvLmx0mamn2WXmrumvsCtoK%2Bd