The coming productivity disruption of generative AI

It surprises me that as recently as December, Generative AI wasn’t even on my radar. Now, it's what most of us who focus on technology talk about all the time.

Things are changing fast in tech. I’ve seen a number of arguments that we are overhyping generative AI, but then I came across this video from Wharton University's Ethan Mollick. (It’s around 10 minutes long and well worth watching.) Not only does it show what these new tools can do, but stresses that companies using the tools have reported between a 30% and 80% increase in productivity as a result. 

The video begins with Mollick introducing his talk, which he created by putting a still picture into one of the tools and telling it to write a script. The result was impressive. While he demonstrates several tools, what is most striking is not only how new the tools are, but how capable they are. Mollick’s well-founded conclusion: not only are we not over-hyping this technology, we’re under-representing how disruptive it will be. 

Generative AI is about to have a major impact on productivity.

Generative AI and app creation

One of the big problems when working with IT or a contractor to create an application is that communication between the parties tends to be suboptimal. This is typically due to several things. One is that the folks asking for an app don’t know how to structure their request so that it’s understandable to those doing the work; often, the request doesn’t include all the dependencies that need to be addressed if the application is to be implemented successfully.

Another problem: the people doing code generally have no idea about the environment in which the application will be used. They don’t know the company culture culture and rarely meet those who will use the application. 

As a result, most of the apps I see created don’t do what the users want. 

With generative AI, the people who need an app can generate the code themselves. And if they aren’t able to articulate what they want to another human, their ability to direct a faster, non-human AI is going to be inadequate, too. The AI tool will respond to what it thinks it was asked and, unless it has some kind of unique filter, should spit out a result faster. But it likely will be worse than if a human had tackled the job. 

One big difference in this scenario is that the user can then iterate the code and have the tool correct the problems created by the initial inadequate directions. The AI tool won’t, as a human might, get upset about the iteration, and can implement any needed fixes quickly. The resulting app is likely to be better, and created more quickly, than if the requestor had gone to a human in the first place. 

More broadly speaking, for real productivity benefits that are closer to 80% than 30%, the requestor will need to be trained to better understand and more completely articulate what they’re asking for. In time, that should massively reduce the number of iterations needed to get to a satisfactory result.

That kind of disruption will play out in different ways across different industries, but it's coming.

Faster and faster?

I agree with Mollick that generative AI is a bigger deal than we think, which is amazing given how much we seem to be over-hyping it now. This technology is advancing at a blindingly fast pace, largely because the folks creating it are also using it in the creation process. And as the tools improve, they will further supercharge related development efforts to a degree it isn’t hard to imagine coders being taken out of the loop and AIs begin evolving autonomously. When that happens, this already fast development process will seem slow by comparison.

The people using these tools still need to be able to communicate with them effectively to gain the full benefits. Clearly, those who can will be in high demand, while those who don’t may find themselves unemployable in markets where these skills are in increasingly needed. (They’ll be just like those who couldn’t use PCs years ago.)

To sum it up, the impact of generative AI won’t just affect the number and types of jobs out there (editors may no longer need writers and low-level coders might become practically extinct, for instance). It will also require users to come up to speed on how to best communicate with these systems or face becoming redundant as well.

Generative AI will eventually impact everyone. Best begin to learn to communicate better now, or you’ll likely regret the impact it has on you.