AI and productivity

It's smart to exercise extreme caution around the use of #genai in the enterprise but it's not sexy and so it's not happening very much. Of course AI has a critical role to play now and into the future for business - I've been an advocate for a decade. And it is true that the companies that figure out how to systematically incorporate it will accelerate their growth much faster than their competitors. And an important reason why GenAI has found its moment is its uncanny way of stimulating our imaginations. That's because of the way we interact with it in a conversational way and in its human-like ability to offer up ideas based (unhuman-like) on the mountains of training data we could never consume on our own, even when there is concern that the training data will begin to eat its own tail. So, why exercise caution? Why the rain on the parade? Because, for now, it probably costs more than it's worth when you factor in employee time and productivity and their well-being.

Probably because they fear being left behind, most companies are haphazardly adopting Generative AI. They're encouraging or requiring employees to utilize it in their jobs to save time. Which on its own is not a bad thing but they aren't doing much in the way of strategic oversight and the provision of guidance either. Because of such a loose approach, the tech is not yet offering the productivity benefits that so many AI leaders and AI companies say it is. Many of us who work with clients see this gap and this research from Upwork puts data behind the argument. One thing that the global survey of 2500 leaders shows is that productivity claims are overblown and, in fact, the case can be made that AI has been counter-productive for employees. Here's an extract from the article.

How can productivity be improved without strategy? As it turns out, as with any kind of business change, good leadership matters. So does structure. And in the case of AI, structure matters if a company hopes to reap strong benefits from it. I'm all for experimentation but even then, there needs to be structure. Otherwise it's just an afterschool project.

Chip Huyen, one of my favorite computer scientists and Stanford machine learning instructor (among many other things) posted recently on LinkedIn what I call an example of a governance model for GenAI applications in the enterprise.

Does that diagram spook you? Does it smack you in the face as being too much corporate? If so, then you probably work for an early stage startup that proudly declares that speed is what matters most. Speed does matter until it results in crashes. Enduring growth is enabled by structure and so it's better to be prepared. I'm a fan of thoughtful construction of processes and Chip's proposal makes sense to me.

Someone else who has been thinking of this, and is worth a follow, is my friend Amir Hartman, who publishes the Leadership in the Loop newsletter. He's been doing some heavy thinking about how companies can both exploit GenAI and protect themselves and their employees from its unintended consequences.

All of this is to say, GenAI is offering a great opportunity for companies to truly make great productivity strides. But it won't happen without strong leadership and without strong process minds getting to work. Let's not blow it by behaving like 6 year old children on a soccer field, all chasing a ball with no real purpose.