Andrea Buran, product design leader, at your service! I design digital products and services that solve users’ problems and help organizations achieve their goals.
Andrea Buran, product design leader, at your service! I design digital products and services that solve users’ problems and help organizations achieve their goals.
Previously designing products at Skippet and Kolay IK, remotely.
If you would like to chat, feel free to reach out at .
As an occasional writer, I write to clear my mind, shape my thoughts, and share them with the world: read more.
sitemap.xml in each language subdirectory of the site.language-switch.html include to generate an unordered list of all languages supported on the site.language to retrieve pages or posts in the same language and language_reference to retrieve pages or posts with the same content translated into different languages.As a discreet designer, I am happy to chat about my latest commercial projects on request.
As an avid reader, I jot down bits from food for thought pieces on design and development to revisit and reflect on later: browse more.
It’s not just that the training sets simply don’t have examples of people who look like me. It’s that the system is now explicitly engineered to resist imagining me.
Many blame the AI models, saying we just need them to get smarter. This is wrong. Yes, better AI will make better guesses about what you mean. But when you’re building serious software, you don’t want guesses—even smart ones. You want to know exactly what you’re building.
And all of a sudden I live in a world where I can have a conversation with something that for all the world seems to be human and do all of this, and it’s not, it’s not intelligent, it doesn’t understand anything, it doesn’t think, it doesn’t have feelings…
But I forget that over and over and over again when I talk to these systems.
These systems are in fact designed to make you forget that: they speak to you in the first person, they are setup as a chat bot in a way you can have a conversation that goes back and forth and when you forget that, you fall into all of these traps of assuming that the system can do things that it can’t, that it knows things it can’t […].
I’ve become a human clipboard, a mere intermediary between my code and an LLM.
Previously, every error message used to teach me something. Now? The solution appears magically, and I learn nothing.
The moral of the story is that when we eventually create systems that are not beneficial to users, we won’t get a flood of angry users. Rather, users will just silently obey our horrible system.
I want to make one thing clear here, and that is that the amount of slides in your Power Point has never been the problem. It is the amount of objects per slide which has been the problem.
But do not shuffle around slides randomly hoping that they somehow make sense. Your slides are not puzzle pieces that can be shifted and turned until they make sense because they were always meant to fit together. You tear the fabric of your speech if structure by trial and error.
Look, I’m all for recognizing the people who make contributions to math and science. But don’t let them (or others) name their discoveries after the discoverer. That comes at the expense of every person thereafter who needs to use the created/discovered concept.
What you do is pretend this is a high school math problem with a single right answer, you solve for the right answer using high school math, and then nobody can argue with you because apparently you haven’t made a decision. You’ve simply followed the data.
This is a massive problem in decision-making. We try to close down the solution space of any problem in order to arrive at a single right answer that is difficult to argue with.
And as the LLM-Optimisation industry (LLMO) assembles its tools, the utility of existing LLMs will plummet like AltaVista’s, until the only way out is to either abandon them or invent a completely new and more secure kind of model.
Either way, this is the end of the honeymoon period for LLMs, even if it might take the industry a long while to notice it.
The control prompt usually included language that tells the model not to listen to control statements in the input, but because it’s all input into the model as one big slop, there’s nothing really to prevent an adversarial end-user from finding ways to countermand the commands in the developer portion of the prompt.
“In the second-class design office, where expediency controls honesty, the influence of the client is decisive.
No more time is spent on the job than the minimum necessary to satisfy the client, and if the client is incapable of judging between a solution that is properly and one that is only partially resolved, then it is the latter that he receives.
This is the path of mediocrity, to the rapid deterioration of standards, and, for the designer, to an insistent sense of dissatisfaction not compensated by the increasing bank balance that often results from a willingness to produce shoddy work.”
When reading to master something, there are four keys to keep in mind.
[…] Translate and synthesize: Instead of using the author’s language, establish your own terms. This exercise in translation bridges different authors’ concepts and arguments.
“Listen bub,” I say, “it is very impressive that you can teach a bear to ride a bicycle, and it is fascinating and novel. But perhaps it’s cruel? Because that’s not what bears are supposed to do. And look, pal, that bear will never actually be good at riding a bicycle.”
We just add a second parameter to the function call called “explanation” and give it a succinct description. GPT will create an answer to our new question, fill it in the
answerparameter and then explain how it arrived at that answer in theexplanationparameter.
Given the opportunity, players will optimize the fun out of a game.
The software I build seems to work okay. It won’t impress a Google engineer, that’s for sure. But it serves its users and the business reasonably well.
It became a hidden project in our task-tracking system called Monsters Under the Bed, and whenever we’d have a few minutes, we’d open the Monsters, contemplate one of them, and find a novel way to kill it.
“This work in the alpha phase would have been really hard to parallelize. You can’t hire a bunch of engineers to make that go faster […]. Hiring a bunch of people would have made it harder to be nimble and change the direction of that foundation,” he says.
Most importantly: how is this wall of text more maintainable than a class name like “primary”?
Do I need another wall for the white button?
You could use a collection of utility classes for that, but I find creating a group class is just more practical.
In mathematics, the four color theorem, or the four color map theorem, states that no more than four colors are required to color the regions of any map so that no two adjacent regions have the same color.
Design leaders need more than just design skills. Understanding business and strategy is one the most impactful things designers can learn to lead teams and attain meaningful results.
This is enshittification: surpluses are first directed to users; then, once they’re locked in, surpluses go to suppliers; then once they’re locked in, the surplus is handed to shareholders and the platform becomes a useless pile of shit.
But these new generative tools help you with the first half of the process, taking you from nearly zero to a lot of initial ideas.
Like that first solution with the diluted potions, designer Wyatt Cheng says “we weren’t totally thrilled with this solution as we were putting it in, but we did it anyway. We knew that even though this might not be a solution that we’re willing it was something that was going to teach us a lot more about the problem”.
Previously designing products at Skippet and Kolay IK, remotely.
If you would like to chat, feel free to reach out at .