Hi there, picnickers!

Today’s issue is a sort of continuation from the last one, which I had to cut for length. But it’s a critically important component of the industry meta-narrative (that companies are cutting headcount because they think AI can make up the difference).

The reason that I think managers think they can get away with this is because they have redefined their expectations from individual contributors (NB: there is no such thing as an individual contributor, because we produce value together) to be entirely centered around outputs, because outputs are easy to manage. On top of PMs trending towards mere project managers, our own reporting chains — who are ostensibly supposed to help us level up in our own competence and achievements — have taken on the role of supervisors who just report on what’s been produced.

It’s ironic that these managers don’t understand how easy their job is to automate. Certainly, the job of merely producing outputs that fit requirements is already being automated, with King Studio laying off 200 people who had just helped perfect their level generating tools. The managers signing off on this decision could only have done so if they believed that the outputs of those teams were all that mattered.

This, I think, is the main disconnect between designers and management. Our stakeholders crave certainty from product/design rather than through product/design.

To be a little more clear about this distinction: when roles ask applicants to be “comfortable with ambiguity” what they inevitably mean is ambiguity of requirements. The people responsible for defining what everyone is working towards are abdicating that responsibility, and you must be okay with that.

What is never ambiguous are the output goals. UX must work 2 weeks ahead of dev. Deliverables are due by such-and-such a date, for release by this other date. We don’t understand what we are doing or why, but whatever it is, it needs to be done on time — because that is what our managers chose to focus on.

My entire philosophy of design is that this framing is completely backwards. Design is the process through which those necessary outputs become defined. Until we have done design, we cannot know what the necessary output will be, or when it will arrive. And when we cede the responsibility to do that definition, we have already lost the plot.

Unfortunately, I think a lot of designers are just fine with that, because outputs measured by velocity rather than purpose mean that you can get away with shipping a load of rubbish.

Which is why so many design leaders have been optimizing for the displacement of responsibility for why, so that they could focus entirely onto managing when.

And that is exactly the mindset that got partners thinking — why should we wait for when to be later when we can have it now? The PM in that link is far from the only person thinking the obvious thought: that if the value the company gets out of design is outputs produced to spec, then an automaton can provide that same value more quickly.

Let’s ignore for a minute that by admitting that their entire value comes from prompting, those same people are signaling that their own work is worthless and focus on the biggest problem with skipping the entire design process to generating cool-looking shit: you are destroying your own ability to recognize what “good” means.

Our field is no different from, for example, infosec — where not a single vulnerability report found with the “help” of AI tools was legitimate. Bug bounty programs are considering shutting down because they are being flooded by well-meaning contributors who were misled by chatbots to think they found something real.

UX is no different. Non-experts do not feel shame at shipping garbage, but appealing to “taste” as a differentiator will get you laughed out of the room by stakeholders who don’t share your set of values. To them, the algorithm can produce something that (in the absence of meaningful, data-informed objections) is satisfactory and therefore fine to ship.

The result is not only that we create terrible experiences, but that those experiences solve the easiest problems, rather than the most valuable problems. Users aren’t buying what we’re selling — but VCs are, so here’s yet another cohort of tools that no one would be using if it wasn’t being made unavoidable.

It’s not that AI is fake and can’t produce outputs. It’s just that by appealing to outputs as the value of our labor for decades, we have undermined our own case for why we shouldn’t be automated away. These tools will continue to eat away at the middle range of quality, producing an extremely small market of bespoke, high quality goods and an extremely large market for people who have to put up with garbage because there is no alternative (and if you’re putting all your chips on getting to produce for that first market — so is everyone else, good luck).

Doing It Right

(Analogous to the section about nice stuff introduced last week, I thought it would be nice to formalize a section about actionable advice that you could take away and do something with.)

There has always been a thread through UX design that has maintained we ought to never have strayed away from our webmaster roots; that we won’t be taken seriously unless we can attend to the material in which we produce outputs. While there is a minimal level of understanding how things work, I think that it not quite what we are missing.

Our first goal should not be functional outputs, but conceptual integrity. Instead of constantly trying to expand the scope of work designers take responsibility for, we should find ways to take the scenic route towards our goal. Harness the divergent thinking of unaligned stakeholders to move everyone roughly in the same direction and you’ll spend less time on getting to a place everyone can agree is valuable.

— Pavel at the Product Picnic