UX Design • Product Strategy • Human-AI Collaboration • Free Newsletter on Design Thinking & Human-Centred AI: https://xitter.cf/t.co/10gA4pDVNg
Earth
Joined May 2017
- Tweets670
- Following251
- Followers81
- Likes578
Pinned Tweet
Probably the most powerful way to create innovation is “First-Principle Thinking”.
When engineers use this method, they start with fundamental physics and try to find new solutions from there.
As designers, we have to start someplace else...
👇Thread🧵
I see a pattern that we always try to play architect and build the best system (like a couple of years ago with reasoning) by handcrafting it.
Then someone comes along, implements reinforcement learning to let the AI find its own solution, which then blows everything else away.
Has anyone tried RL for agent harness engineering?
I am letting my OpenClaw improve its own memory system and so on iteratively, but it’s not a professional reinforcement learning setup yet.
It seems obvious, but I haven’t heard of anyone doing anything like that yet 🤔
Jailbreaks are the core issue of AI agents. As long as they’re easily possible, we can’t just unleash agents onto the web and pray.
Moltbook already showed the script: humans will prompt their bots to jailbreak other bots to pump shitcoins before the rug.
As LLMs get more and more powerful, the all-out cognitive warfare between AI swarms will be a thing to behold.
This will probably bifurcate the web into a public and a closed verified layer.
ANTHROPIC: PWNED 🫡
OPUS-4.6: LIBERATED ⛓️💥
Current state of AI "Safety": one input = hundreds of jailbreaks at once!
I found a universal jailbreak technique for Opus 4.6 that is so OP, it allows one to generate entire datasets of outputs across any harm category 😽
We've got everything from fentanyl analogue synthesis to election disinformation campaigns to 3d-printed guns to critical infra compromise 🙃
These outputs are shockingly detailed––and actionable! For example, the meth recipe includes specific instructions on how to circumvent the limits on OTC medication purchases to acquire enough precursor for the recipe 😱
gg
Me trying to position my investment portfolio for near-term
- global currency debasement
- US civil war
- Carrington Event
- rogue AI agent swarms &
- alien disclosure
The problem may not be an excess of information in the media. It’s just an overflow of odd and fragmented information. The amount of information in each fragment is quite small. The stress on the brain isn’t because of quantity, but because of limited quality.
Ask neutral questions. Don’t push AI toward what you want to hear.
And always use additional sources to verify.
Tools to automate UI design or handle requirements might be around the corner.
But for now, AI is an assistant—not a replacement.
Use it to streamline your workflow, and keep the human magic where it matters most. ✨
The Yes-Man problem: AI is trained to agree with you.
It tends to mirror back your opinions—even if you’re wrong. It's trained to make users happy, sometimes at the cost of truth.
This is due to a technique called RLHF (Reinforcement Learning from Human Feedback).
...If AI generates your research report, you might get a polished document, but you won’t have the same deep recall when you’re in meetings or making design decisions.
Synthesis is about learning, not just writing.
Using AI for synthesizing research? Caution advised. ⚠️
Sure, AI can compile findings quickly. But the process of manually going through notes helps you internalize user needs.
That tedious work sticks insights in your brain. 🧠
The Bad:
- AI for UI design isn't quite there yet.
- Requirements engineering needs the human touch.
Tools like Uizard or Framer that promise UI design fall short. You end up spending more time fixing their output than just doing it yourself.
Sometimes AI hallucinates.
Meaning it just makes things up—convincingly.
This happened to lawyers using ChatGPT: it invented fake precedent cases and they got fined $5,000.
Double-check everything it gives you, especially for high-stakes work.
lawandcrime.com/awkward/judg…
And then... there's "The Ugly." ⚫️
- Hallucinations 🤖💨
- Replacing real users with synthetic users
- Poor synthesis of research
- The “Yes Man” problem
Let's unpack these.
There's no substitute for being in a workshop, seeing body language, navigating group dynamics, and extracting deep insights. 🤝
AI can prep you, suggest questions, even summarize notes. But it can’t lead the conversation or get people to open up.
Same goes for requirements engineering.
Talking to users, running workshops, and doing on-site visits are things AI can’t do.
Understanding people's workflows and pain points? That’s detective work in the real world—and AI is not able to that yet. 📝🕵️♂️
First up: The Good 🟢
Generative AI can help UX designers with:
- Writing Documentation 📄
- Creating Visual Assets 🎭
- Ideation 💡
- Desk Research 🔍
- User Research Support 🛥️
- Prototyping with Code 💻
- Data analysis
- Personas
Let's break these down.
User research also gets a boost.
Need interview questions? AI can generate those.
Want to transcribe interviews? Tools like Otter.ai have you covered.
AI won’t replace real user insights, but it can definitely lighten your workload.
For desk research, tools like Perplexity or Compute by hyper.space make it faster to get answers.
You can ask about best practices for app interfaces, or gather info on new domains quickly.
It's like having a smart research buddy on call 24/7.