@scottjenson@social.coop
@scottjenson@social.coop avatar

scottjenson

@scottjenson@social.coop

UX Strategy: Apple System 7, Newton, and Apple Human Interface guidelines. UX Director at Symbian, manager Mobile UX at Google, creative director frog design San Francisco. Head of Product for two startups. Returned to Google working on the Physical Web project and Android UX research. Now semi-retired

mastodon.social: 2017-2022

This profile is from a federated server and may be incomplete. Browse more on the original instance.

bruces, to random
@bruces@mastodon.social avatar

*I bet it wasn't all that boring if you were a Google AI and you were just automatically scanning it

https://gizmodo.com/google-io-boring-gemini-ai-1851485806

scottjenson,
@scottjenson@social.coop avatar

@bruces I just left Google, they have AI fever. This is NOT something driven by a user need. It is a stone cold panic that they are getting left behind.

The vision is that there will be an Ironman Jarvis in your phone that locks you into their ecosystem so hard, you'll never leave. That is just catnip and they can't afford to let someone else get there first. (Apple is no different)

When the emperor, eventually, has no clothes, they'll be lapped by someone thinking bigger.

scottjenson, to LLMs
@scottjenson@social.coop avatar

Saying "LLMs will eventually do every job" is a bit like:

  1. Seeing Wifi wireless data
  2. Then predicting "Wireless" Power saws (no electrical cord or battery) are just around the corner

It's a misapplication of the tech. You need to understand how work and extrapolate that capability. It's all text people. Summarizing, collating, template matching. All fair game. But stray outside of that box and things get much harder.

scottjenson,
@scottjenson@social.coop avatar

@mattwilcox Oh, I agree. But there will be domains where it's transformative, e.g. programming. This has completely transformed how I code. I'm still driving, but DAMN is it handy.

I expect it'll transform online help significantly (I hope for the better!) It will likely transform Law in many ways. I don't expect it'll replace lawyers, just allow them to review and output boilerplate much faster.

It's the Gartner hype cycle all over again. People are freaking out and it'll pull back.

scottjenson, to Figma
@scottjenson@social.coop avatar

I just tried a few AI plugins for and they were all bad. This domain might be a great test for . I predict these failings are unlikely to be fixed any time soon:

  • Layout was poor
  • They can't create components
  • Laughably complex object hierarchies (everything was enclosed in a frame)

Of course things will improve, but I expect fixing these deep structural problems are a function of many new constraints, likely beyond what today's LLMs are actually capable of. @simon ?

scottjenson,
@scottjenson@social.coop avatar

@simon my point being there are limits as to what #LLMs can do:

Structural
There is no clear API to "genAI" components

Training
There is very little training data on how to create a clean Figma object structure

These may be solved, eventually, but they also are likely quite different from the chat based solution patterns offered today. My concern is that it's much harder than boosters believe.

davew, to random
@davew@mastodon.social avatar

Do people understand the impact ChatGPT can have on how we control our software? No more hunting through menus among a thousand options to figure out how to do something. Just use natural human language. The UIs we've had to design and live with will be a thing of the past. Yet no one seems to be talking about this.

scottjenson,
@scottjenson@social.coop avatar

@davew People in the UX community are most certainly talking about this. There is tremendous potential here. There is a tendency though for more technical users to over-index on typing as a perfect experience. For repetitive tasks, it can get old. This doesn't take anything from you point. To the contrary, it's WHY we're talking so much about this.

Strapping an LLM powered CLI into an app is an obvious first step but we'll likely do something much more interesting

scottjenson,
@scottjenson@social.coop avatar

@chrisgervais @davew exactly. that type of "let me describe it and it'll do all the work for me" type front end seems like a very interesting direction to go. The gotcha, however, is how often is it going to get it wrong? Even describing something to another human is error prone!

matthiasott, to UX
@matthiasott@mastodon.social avatar

A little quiz: in which of those two inputs will your password be visible aka unmasked? 🤔

scottjenson,
@scottjenson@social.coop avatar

@matthiasott This is a UX problem as old as time. It's why checkboxes were invented. They are a label AND a state! I understand there is a need for minimalism but it's basic UX Algebra: an icon can be either an action OR a state not both.

The solution is to make it a button and 'sneak in' state someone, e.g. grey prompt text in the text area that toggles with the icon (there are likely better ways to do it)

scottjenson, to random
@scottjenson@social.coop avatar

@simon
Simon, I'm working with @homeassistant a bit and we just had a fascinating discussion about 'nanoLLMs' that could run locally. They would NOT need the sum-total-of-all-human-knowledge but would really just be there as a smart parser for speech-to-text commands, keeping everything local. This is clearly still not trivial but hopefully one way to reduce the model size.

Do you know of any 'reduced' LLMs that could work in this more limited context?

scottjenson,
@scottjenson@social.coop avatar

@simon @homeassistant Excellent news, thank you. I'll get started running it locally on my Mac just to get started

scottjenson,
@scottjenson@social.coop avatar

@simon @homeassistant Thanks again. I'll also give that a try. I'm currently running it with Ollama on a 5 year old desktop (it was shockingly easy) It's only using 30% of the CPU when I ask it a question!

Even then, I'd (naively?) suggest that it is far more power than I need. But the fact that I can get this far in just 5 minutes has me shaking my head in disbelief.

scottjenson, to ChatGPT
@scottjenson@social.coop avatar

My experimentation with #ChatGPT #Gemini + programming has been a mixed bag:
Pros:

  • It often gets it right the first try
  • It's faster to change the prompt than edit the code
  • it can explain code fairly well

Cons

  • I can ask the exact same prompt and get different code that WONT work!
  • After several iterations, the code just stops compiling
  • I frequently save revisions of files so I can upload a working file and iterate from there.
  • It's SO SLOW to wait for it to type out
scottjenson,
@scottjenson@social.coop avatar

I've been wading into code/APIs I have zero experience with and making remarkable progress. I'm thinking of it as creating a good starting tutorial.

I'm still giving it fairly tiny utility programs (I am just prototyping crazy stuff) I'm not building anything complex. But as a Designer the fact that I can build a working prototype in or so damn fast is remarkable.

davew, to random
@davew@mastodon.social avatar

I can’t wait for the UIs of settings on systems like Mac Or Android to go through the AIs. No more hunting through menus to not find the setting where you’re sure it should be.

scottjenson,
@scottjenson@social.coop avatar

@davew MacOS has a nice feature where if you searched in the settings app, it included dozens of synonyms so if you searched for "audio" it would find "sound" etc. An AI interface could do this MUCH better.

However, it doesfeel like were going back to CLIs a bit doesn't it? Having to always type to get anything done? It is, of course, a really clever CLI, but might feel a bit tedious if it's the primary way you need to use it (audio just doesn't work in many so many social situations)

scottjenson, to random
@scottjenson@social.coop avatar

The world needs some cross between Processing, HTML, and Google Docs.

Processing: A programming environment with lots of helpful libraries
HTML: a simple display model that can be viewed everywhere
Docs: an underlying sharing mechanism

There are just too many core problems that need to be solved over and over by each app. It's more than just getting the right library as there are services involved.

Surely someone has taken a crack at this?

scottjenson,
@scottjenson@social.coop avatar

Any yes, HTML/JS/CSS can be used to build anything, but that's a bit like saying Assembly language can build any app: At some point you don't want to keep reinventing the wheel.

Of course, EVERY programming environment has tried this and it's nearly impossible to get something to work universally. I'm just asking for a very opinionated set of services and libraries on top of the web so it's really stupidly easy to build and SHARE things easily.

Kind of like Glitch but with a few more mix-ins

scottjenson,
@scottjenson@social.coop avatar

@maximvdw Very cool. It appears to be very storage focused (which is a great start) but does it have a Glitch-like side that allows for programming that data?

scottjenson, to random
@scottjenson@social.coop avatar

I was 'kicking the tires" signing up for Threads and the VERY FIRST thing I noticed is that people I don't follow are showing up in my feed!

After so much time in the fediverse, this just feels like a violation. NOPE NOPE NOPE

scottjenson,
@scottjenson@social.coop avatar

@davew Yeah, I get it, it's the classic problem all social sites have. I don't hold that against them, it's the fact that I cant see only my followers that I have an issue with. Apparently the native app allows this (just not the web version)

scottjenson,
@scottjenson@social.coop avatar

@ramsey @peterme

I

scottjenson, to UX
@scottjenson@social.coop avatar

For all of my complaining about the of kitchen appliances, my new Breville oven is actually astoundingly good.

Not only are the controls clear and easy to use but there are small little touches that are just very well thought out. For example when the timer gets close to zero the internal light turns on automatically.

craiggrannell, to random
@craiggrannell@mastodon.social avatar

Looks like the UK weather has finally found the heating dial.

(For US folks, the high range here is approx 68–79 Farbalinoos.)

scottjenson,
@scottjenson@social.coop avatar

@craiggrannell It's not just a math issue but also cultural (e.g. 28C isn't really the same as 82F). When I was in London years ago, there was a tube ad for air conditioning saying "No one can work when it's 24" (with an image of a sweltering office worker)

That was funny to me as 24 to most yanks is near perfect

scottjenson, to random
@scottjenson@social.coop avatar

It's clear the "Turing Test" is failing, just as it failed with the Eliza chatbot back in the 1960s. People are just too easily fooled.

For me the REAL test occurs when an "AI" is NOT speaking to you. What does it do with it's free time?

https://en.wikipedia.org/wiki/ELIZA

scottjenson,
@scottjenson@social.coop avatar

Before anyone chimes in, the answer is absolutely zero. These things ONLY do things when prompted. They basically are entirely Parrots. There is ZERO concept of awareness or even goals. That may come of course, but my point is that, at the moment, it's not even a glimmer.

scottjenson, to UX
@scottjenson@social.coop avatar

There is so much bad #UX #design here that I hardly know where to start.

If you are a professor, this is an excellent design exercise: improve this. I'm sure ANY student could make this far simpler and easier to use.

I have a short video using this that makes it ever more clear this was designed by a dysfunctional design team. This is a #Dodge #Durango if you're curious.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • provamag3
  • InstantRegret
  • magazineikmin
  • modclub
  • khanakhh
  • Youngstown
  • rosin
  • mdbf
  • slotface
  • Durango
  • ngwrru68w68
  • thenastyranch
  • kavyap
  • DreamBathrooms
  • JUstTest
  • cubers
  • osvaldo12
  • Leos
  • anitta
  • everett
  • ethstaker
  • GTA5RPClips
  • tester
  • cisconetworking
  • megavids
  • tacticalgear
  • normalnudes
  • lostlight
  • All magazines
    •