Michael Watkins: You can’t afford to be a dinosaur

One of the things that's just so fascinating about AI is how rapidly things can change

At the start of 2025, the major players, OpenAI, Microsoft, developing these new models

thinking they're ahead of everybody in the world. And then, bam, this little startup from China called DeepSeek introduces a new model

as good or better than any of the existing models out there. and in a day a trillion dollars of valuation of those companies is gone. Yeah John we've got a bit

of a tech sell-off this morning and it's being caused by earth-shattering developments in the AI

space. An enormous uncertainty about you know the progression and profitability of the industry rises

and so I think it's just an illustration of how rapidly the landscape can just shift in very

fundamental ways and we have every reason to expect that that's going to continue to happen

When I talk about AI, I often put up two slides. The first is the dinosaur admiring an incoming asteroid

And the second is a surfer surfing a very large wave. You cannot afford to be the dinosaur

You must be the surfer. You need to be agile. You need to be adaptive

You need to embrace these systems and try out what's going on, because if you don't understand it

you really can't hope to help your organization adapt and stay in front of this momentous, world-changing technology

I'm Michael Watkins. I'm a professor of leadership and organizational change at the IMD Business School

I write extensively about leadership, about strategic thinking, and about the impact of AI on both

so of course in november of 2022 in generative ai exploded on the scene but i think it's important

to realize that artificial intelligence has been around for a lot longer and has gone through a

very interesting and important evolution previously artificial intelligence was mostly about ytics machine learning pattern recognition Largely that been kind of invisible in some sense to most people except in how they use the tools

like the kinds of things that power your Netflix recommendations or your Amazon selections

But of course, then generative AI really changed everything. And generative AI, as the name indicates, is about generation

It's about content generation, it's about creativity, it's about writing, and it's really changed pretty fundamentally how I think virtually everybody doing white-collar jobs is doing their work these days

And of course, AI continues to evolve. The most recent models are known as reasoning models

As the name indicates, these models are capable of what's known as chain-of-thought thinking

They're reasoning their way through problems step by step in a way that previous generations of LLM simply were not able to do

And that has enormous benefits for things like solving problems in science and mathematics

Planning things becomes possible in a way it wasn't before. And this is, again, going to generate an enormous change in how the technology is used in business and beyond

So now the cutting edge is really evolving towards what's known as agentic AI

And that's AI as agents, right? Agents out in the world doing things that are important for us

That could be, you know, booking your next flight for your vacation or deciding what you need in your refrigerator to feed yourself for the next couple weeks

And this is really an important evolution because it begins to make AI very active in the world

in ways that are going to both, you know, generate tremendous benefit, but also potentially create some significant risks

As soon as you see AI beginning to be able to think and reason and then begin to make recommendations and decisions based on that thinking and reasoning you open up the possibility for serious misuse serious misguidance potential for misinformation even the ability of these systems to influence us in ways that we don fully yet appreciate or understand Today the way we interact with AI mostly is through what known as prompting which

essentially boils down to asking the AI, either in writing or verbally, to answer certain

kinds of questions for us. As you're learning to prompt these systems, there's a few limitations and watchouts you

should be keeping in mind. The first is what's known as the hallucination problem, which is the tendency for these systems

if they don't have an answer, not to say, hey, I don't know, but instead to give you

some fact that turns out to be completely erroneous. When I started using these systems

to write articles, for example, I would ask it for some references, and it would very confidently

give me three or four references, which when I checked, did not exist. Now, you can trust their

creativity, you can trust their ability to help you generate ideas, but when it comes to facts

you are well advised, very well advised to be double checking almost anything

A second watch out is these systems are programmed with personalities that are intended to be

helpful. They're going to tell you nice things. They're going to be complimentary to you

They want to make us happy. And you've almost got to work kind of a little bit hard to say

don't try to make me happy. And you can imagine the vulnerabilities that that raises, right, for

all kinds of high criticality applications where we need to be pretty certain that the right kinds

of things are happening. And then that leads to a third big watch out, which is they seem to have

a programmed desire to think that everything is going to be okay. I was writing an article recently

about the potential employment impacts of artificial intelligence. And I asked it a question

What do you think is going to happen with employment? Oh, it'll probably be okay if we do the following things

And then I said, well, no, wait a minute. Are you being over optimistic? And if so, stop

Well it going to be a disaster There going to be probably you know loss of three or four jobs for every job that created And so understanding that there is this seeming inherent bias in these systems towards a kind of optimistic set of outcomes especially regarding the impact of AI on the world

And again, you've got to be very precise and very thoughtful to ask the system directly to de-bias itself, to give you the real truth as it sees it

And even then, you can't necessarily be sure that that's absolutely what you're going to get

As you think about learning to be a great prompter of AI, it's a bit counterintuitive because it's quite different often than how you would interact with another human being

AI has no ability to understand the broader context in which the question is being asked

It has no ability to interpret the emotional resonances that you're trying to communicate

This is particularly true with the reasoning models because you're asking them to go through a chain of thinking

And if you start at the wrong point, it's rapidly going to head off in directions that are not going to be all that useful

So especially for the reasoning models, giving it lots of context, lots of specificity, being as precise as you possibly can from the outset is absolutely the way to go

You personally need to engage actively with this technology. And it's not enough that you just have some basic literacy

You need to be experimenting with new models as they come out. Because only by doing that are you going to learn what the capability is

and begin to track what's happening as the capability continues to advance

We have to be vigilant about the potential downsides, and we have to be rapid to embrace the upsides

And above all, you have to be the surfer, surfing the wave as things go forward

It's the only answer that I've come up with for how to continue to thrive in a time of really extraordinary change

Get smarter, faster with videos from the world's biggest thinkers. To learn even more from the world's biggest thinkers, get Big Think Plus for your business