# Strategic Uncertainty: Scenario Planning vs. Future Prediction in AI

**Podcast:** All Things Product with Teresa and Petra
**Published:** 2026-04-21

## Transcript

Hi, folks.
This is All Things Product with Petra Wille.
And Teresa Torres.
And we're so happy you're here.
Teresa, I heard that you think that people should stop to predict the future.
Help us to understand what that is all about.
Yeah, we are recording.
A couple, like a month or so before this is going to go live.
So I'm just going to start with what's happening right now as context for this conversation.
So there's been a number of articles that have gone viral and like massively viral around all jobs are ending.
The economy is going to collapse.
That's like one side of it.
There's been a lot of people predicting what skills are going to be most important in a world of AI.
There's a lot of fear mongering.
That's the part that really bothers me.
But there's also this certainty behind these arguments.
This is what will happen.
And I think this is the part that ruffles my feathers a little bit.
We have a lot of research on predictions and experts are terrible at predictions.
Novices are certainly not good at predictions.
We're just not very good at predicting.
And so when I read one of these arguments, no matter what the topic is, where they're just a hundred percent confident, this is what will happen in the future.
To me, it's like this blazing, like flashing neon light that's saying I'm uncomfortable with uncertainty, right?
Like that's really what's happening.
I'm uncomfortable with being a human being because then you're always exposed to uncertainty.
Yeah.
And so I wanted to just talk through that a little bit.
Like we are going through a period of change.
I have a feeling it's going to continue to accelerate, especially the impact of AI on the tech industry.
Again, for context, we're starting to see this with software engineering, where a lot of engineers are starting to share.
They don't write code anymore.
This is going to be the end of software engineering jobs.
And like, again, I know better than to predict the future.
That's the whole point of this topic.
But we can look at like what's already happening right now at the leading companies.
They're not firing their engineers.
They're doing more work, right?
And so it's really easy to get caught up in this like reading what seems like a very rational argument about what's going to happen tomorrow or next year or five years from now or 10 years from now and like jump and like just fall prey to it.
But I think like I always go into those articles with this headline, literally flashing neon sign of the author saying I'm uncomfortable with uncertainty.
But I love that you're still reading them, Teresa.
I love that you're still reading all these posts.
Okay, this is because I think it's still important.
Like, here's what I don't like.
I don't like, I think when we try to predict the future and have certainty around it, we're not acknowledging the ambiguity and uncertainty in the world.
But what I think is useful is scenario planning.
Yeah, yeah.
OK, yeah, I agree.
Lots of different potential futures and then going through the exercise of like, if this happened, what would I do?
Like, that's a very productive exercise.
But that's not what these articles do.
They don't write about like, here's 12 potential futures.
What if we did these things?
They write about one doom and gloom future or they write about one glorified, amazing utopian future.
And like, really, the more productive exercise is.
Here's a range of things that could happen in the future.
How would we react?
How would we respond?
What should we be prepared for?
Yeah.
And I think we hinted on that in several of our other episodes.
What I think is really helpful, at least to me, is this kind of yesterday's weather planning to some extent, as we always do in Agile, right?
So I just look back at...
my career in software engineering or software development.
And then I just reflect on the patterns that I see.
And we had that multiple times, right?
So API is revolutionizing everything.
Basically, everybody will be unemployed tomorrow because systems can talk to each other finally.
And same is with mobile apps.
Nobody will need all the web developers anymore because mobile apps is the future and nobody will use a desktop machine.
I'm currently using one while I'm speaking.
So, right, you can look at the past and see that some of the predictions were true to some extent.
And it is a bit of a corridor.
And I like that you said, like, you still read the extreme posts at the end of this corridor to see where things may be headed.
But there is a lot of truth in between these extremes, I think.
Because, for example, are we seeing engineers?
doing a lot more with these AI tools.
Yes, kind of a few of them, the rock stars that we always see doing amazing stuff with new technologies.
But as always, they're early adapters.
And then they're all the laggards.
And that's the same thing with software engineers as it is with every group of in our society.
There are people that are already unlocking a lot of the potential of the AI tools that they're currently having.
But let's face it, a lot of software engineers, a lot of engineering teams, a lot of product people, a lot of organizations will need a bit of time to get there.
Yeah.
Okay.
I think there's two parts to this.
I think there's, it is true.
We can look at like.
companies, people that are living on the edge of like what's possible right now and get a glimpse of what the future might hold.
Like there is some truth to that statement.
So like if I take a company that's fully AI native, their engineers have completely embraced coding agents.
They're no longer writing code themselves.
Like that is a potential glimpse of the future of like, okay, well they're not firing their engineers.
They're actually just building 20, 20 X.
10x, 5x, whatever multiplier you want to put on it that they were doing in the path before.
But here's the thing.
Those people are not just early adopters.
They're also outliers.
And we know from the adoption curve that we don't always cross the chasm.
So it's not true that that is everybody's future.
It's a potential future.
So like I'll use my own experience with Cloud Code.
Like I've been nerding out with Cloud Code and it's completely changed the way I work.
Do I think this is the way all people are going to work?
I think there's elements of it that might be the way all people work in the future.
I actually don't think all people are going to work out of the terminal.
That seems insane to me.
I don't think all people are going to build all of their own software and never buy software again.
Right.
And so like the hard part is it feels like we can look at these leading companies or our own edge experience and be like, this is the future.
What we don't know is what are the elements that will survive as it crosses the chasm to the laggards?
And what are the elements that are going to completely go away?
We have no idea.
And it might change something like a make or buy decision forever, but there still will be make or buy decisions to make, right?
And I think that applies for so many aspects of the...
engineering job of the design job of the user researcher job of the product job basically all of our jobs and i know they're coming closer together there might be more of these product builder personalities i i get all of that but still you will have the conversations that we're currently having and they will be different but they still need to happen here's There's a logical mistake I think people make in these predictions.
I'm going to use one example of an article I read to try to expose this.
So one guy, he's probably what we're going to call on this leading edge.
He wrote an article about how he no longer uses any GUIs.
He set up MCP servers for every tool that he uses in his business, and he uses Cloud Code as his interface.
And he talked about the value of this is he can say, Claude, what's going on with this customer?
And what Claude will do is it will go to his accounting software and tell him if it has any, if that customer has any unpaid invoices.
It'll go to his support tickets and aggregate any challenges.
It'll go to his analytics and see how the, right?
And let like Claude can give him this synthesized view that no individual tool can give him.
Yep.
Okay.
That is amazing.
And I love that.
And this is a lot of the reasons why I use Cloud Code.
There is this element of like the piece that I would pull out of that story is it is becoming easier and easier in software to create custom views.
I don't think the element from that story is people will never use a GUI again.
The interface will be Cloud Code for everybody.
Like that's the like the tinker early adopter thing.
What I think is way more likely is a company or these companies will build connectors and there will still be a battle for who's going to create that UI for you.
Because most people don't want to do that synthesis.
They don't want to connect all those connectors.
They don't want to think about what's all the data I need to know about this customer.
They hire a product to do that for them.
And there is still...
value in for example visual hierarchy to things because you and i using cloud code a lot and you reading text a lot from morning to sunset basically and i think that's not everybody's cup of tea sometimes you need it in a nicer visual representation you need to see things on the first side so all of that what design did the last 100 thousand years basically for us um since people draw in caves i guess um is that they thought about information hierarchy and all these kind of stuff can you brief growth code to do this with your his responses yes to some extent but still sometimes a nicely designed gui will will make the cut and on and it's such a It's such a niche perspective and it's not inclusive and it's not diverse.
And we have all the people that are maybe not good at reading.
So how would they access all the data?
And yada, yada, yada, right?
So all the conversations that we were having the last 100 million years don't go away.
But I think it's an interesting case to think about and to explore the extremes of such a technology.
And I think for product people, it's super important to think about, ooh.
How could that change the UIs that we're currently building?
Because maybe they're more use once, throw away, create a new one when you need a new one or something like that.
Maybe stuff is not as persistent as it currently is.
Maybe views are not here to say for the next five years in your software, most likely, by the way.
I think it still is important to think about how that might change the work that you do and how you assess stuff and maybe what impact it has on your product principles or design principles.
You need to review those things.
But yeah, it's not a good prediction for the future.
I think the logical fallacy is my experience will be everybody else's experience.
And everybody will like my experience.
Exactly.
Like what works for me will work for everybody else.
I do think there are takeaways from these stories, just like you were saying, like how does this affect what we might build?
And to me, this is the scenario planning part.
Like if we were, okay, let's push this to the extreme.
Let's say that all GUIs went away and everybody only interacted with the agent.
What would that mean for our product or service?
Yeah.
I think that's a great thought exercise to go through.
I wouldn't predict that's what's going to happen, but I would scenario plan around that and play with that and see.
Because the advantage of these scenario planning sessions is not just to be prepared for a wide variety of outcomes.
When we push scenarios to the extreme, they actually give us different lenses in which we can view what we're building and who we're building for.
And it allows us to explore more of the problem space and more of the potential solution space.
And that helps us build better products right now, regardless of what happens in the future.
So I think there's huge value in exploring these different perspectives.
I think where we get into trouble is locking onto one as if it's a certain truth.
Yeah.
And it's always, so for me, I get a lot of this tech spike vibes.
So ages ago, engineers came out and said like, Petra, we need a tech spike.
There is this new technology.
Can we have like half of a sprint of our time?
We want to play with this new thing.
We think it could be do amazing in our product.
And if you said like, yeah, please do so.
I can see your point.
Let's see where this takes us.
Then they experimented for a sprint.
The prototype looked amazing.
It really felt good.
But then you have to bring in non-functional requirements.
So is this maintainable?
Is it enhanceable?
Is it actually having the most basic security measures?
Is it taking care of data privacy?
And all these kinds of things.
And with all these stories, I always have this kind of, okay, but nobody brought in the non-functional requirements.
So this is kind of the vibe that I get from these really extreme stories that we're seeing right now.
Okay, so I think to try to make this actionable, I want to think about like, I think people are not going to stop predicting the future.
So I want to think through like, when you're reading an article where somebody's predicting the future, like what are some things that you can do?
What are some things that you can do to like, help evaluate that as one of many scenarios.
And so I think the first thing is like, be really aware of this logical fallacy of like, people are going to write from the perspective of my, what my experience will someday be everybody's experience.
I think it's a really common one.
I think even just framing it as this is one potential future, what are other potential futures?
And like, instead of trying to predict the future, play with a range of potential futures.
And then I think there's this third piece that you were starting to get at, which is like, how could this inform my work?
Like, how do I use this as a lens to change the way I think about what I'm doing?
And like that story that I was telling about this guy who like connected together three things, like the way I would use that to inform my work is what he did was he connected products to solve more of a whole.
problem, right?
Like those products each solved a slice of a problem and he wired them together to solve a whole problem.
And I think that's a good takeaway for product teams.
How do we move our product towards solving more of a whole problem?
So you were trying to find the meta, the meta layer to what you actually were reading.
And then you try to use that as a lens for your work.
Yeah.
Yeah.
And I like your scenario.
I like your scenario advice.
especially for people in product leadership right now, I think it does make sense to have various scenarios at a super messy whiteboard somewhere.
You can pin all these articles and maybe it doesn't make sense in the beginning, but at some point there will be a few scenarios manifesting and then you can think about them and explore them.
Maybe even you will see spots where you have a lot of content.
So that's maybe a strong primer that it is a more likely scenario.
I think that collect some of the content and the bits of the pieces and see if they're pattern.
And I like this idea of the scenario planning, explore scenarios.
Yeah.
I think it helps, um, uh, like cut the edge on the like fear mongering, right?
Like, okay.
So this is, you know, like one is all junior level jobs are going to go away.
Okay.
That's one potential thing.
It could happen.
Right.
But what could also happen is that like, AI could help train junior people so they come in as mid-level people.
That could happen, right?
And so like just that exercise of like, well, what else could happen?
What are the other things?
That's another important one.
To just cut the edge.
There's so much fear-mongering around these stupid predictions.
As always, this was an episode under the headline of Teresa and Petra fighting black and white.
Yes.
Nuance.
Nuance, people.
Nuance.
Thank you, Teresa.
Thanks, Petra.