A.I. Short Takes

To complement Henry’s Short Takes presentation this month, I wanted to bring attention to three fun A.I. image generation apps running on Apple Silicon.

Diffusers from HuggingFace is the simplest of the three.  It’s free in the Mac app store and lets you create images from text prompts. It is less capable than the other apps mentioned here, but by far the simplest to use.

DiffusionBee is a free A.I. app in the Mac App Store that can quickly and easily generate images from text prompts or other images.  It has an easy to learn interface and produces much better pictures than I was able generate using OpenAI’s Dall-e in our SMUG meeting late last year.  DiffusionBee is built on StabilityAI’s Stable Diffusion technology.

Draw Things is another free A.I. app in the App Store, remarkably it also runs on iPhones and iPads.  The UI for Draw Things is considerably more extensive than that for DiffusionBee and it takes a bit more effort to produce a good image, but it can provide spectacular results.

All of these apps can use any of the tens of thousands of Stable Diffusion Models available from HuggingFace.co and CIVITAI.com. CIVITAI also has hundreds of thousands of A.I. generated images. (FULL DISCLOSURE: The site is pretty much uncensored).

Finally, if you don’t have a device with Apple Silicon available, many of the models can run on the HuggingFace website.  For a fun image generator that uses both a text prompt and a starter image, try the InstantID demo on HuggingFace.  Upload a picture of yourself and regenerate it in any of 9 different styles.  Each time you run it you get a new interpretation.

There are lots of other Stable Diffusion UIs around, including those for generating Audio and Video, but the three presented here are my recommendation for a place to start learning about Stable Diffusion image generation.

 

Technical Magic

Artificial Intelligence is the all the rage currently.  People are using online apps such as ChatGPT as though they were actually intelligent, when they often simply assert nonsense.  ChatGPT will, for example, confidently give you instructions on how to drain the radiator in a 1973 VW Superbeetle, a car with an air-cooled engine.

From a start-up named Technical Magic comes Wispy, an AI browser extension that can actually make your browsing experience more productive.

  • Wispy can summarize the web page you’re looking at in prose or poetry.
  • Wispy will proofread a blog post you’re writing, and it can help break even a stubborn case of writer’s block.
  • Wispy can look at a page and separate facts from opinions.
  • Wispy can give you an English text summary of a YouTube video, even one in a foreign language.
  • Wispy will happily suggest side dishes to go with an entree, suggest recipes based on what you tell it is in your fridge, or make suggestions for your next trip to Trader Joe’s.

Wispy has more than 50 “powers” like these and more are being added every day.

Full disclosure:  Wispy was created by friends of mine.

The Technical Magic team is composed largely of former Mozilla employees who have a deep understanding of how browsers work and how they are used.  You can find Wispy at the Technical Magic website.

*************

When I asked Wispy to summarize the blog post above in Limerick form it returned:

There’s an AI named Wispy, you see,
From Technical Magic, it came to be.
It can summarize web pages with flair,
And proofread your blog post with care.
With over 50 powers, it’s quite a spree!

Wispy separates facts from opinions,
And gives YouTube summaries with precision.
It suggests side dishes for your meal,
And even helps with writer’s block ordeal.
A magical browser extension, it’s a true vision!

Created by friends, it’s a delightful creation,
By former Mozilla employees with dedication.
Wispy knows browsers inside and out,
Enhancing your browsing without a doubt.
Find Wispy at Technical Magic’s location!

Jono Xia – These Things I Believe

Re-posted from Jono Xia’s blog: Not the User’s Fault

These things I believe about software development and user-interface design.

1. Why write code?

Software is for humans, not for computers.

Software is only as good as the improvement it makes to a human being’s life.

Are we making someone’s job easier? Letting them have more fun? Helping them learn? Helping them keep in touch with friends and family?

Are we making the world a better place?


2. What do people want?

Most people do not want a computer.

They don’t even want software.

For us software developers, this is a painful truth.

If people don’t want a computer, why do they use one?

  • Email — for writing to other people.
  • Instant messaging — for talking to other people.
  • The web browser — for reading what other people have written.
  • Word processing — for writing something you’re going to print out and show to other people.
  • Graphics — for creating artwork. To show to other people.
  • Presentation — for communicating your brilliant plan. To other people.
  • Games — especially games that you can play online. With other people.
  • Social networking websites — Enough said.

The computer is merely an intermediary. A poor and frustrating one. It is a necessary evil that people put up with in order to get what they want.

What they want is a better way to talk to each other.


3. Why does software succeed or fail?

We software developers, being not exactly social creatures by nature, must work extra hard to understand the social impact our software will have. If the social effect is not what people want, the software goes unused.

We software developers, being not exactly average users, must work extra hard to understand how average users will relate to our software. We see the trees, they see the forest.

We software developers have often been confused and frustrated when a clearly superior technology fails, while a clearly inferior technology spreads like wildfire and takes over the world.

We were surprised because we want each technology to be judged only by its cleverness, its raw power, the cleanliness of its architecture, the purity of its ideas. We were blind to the user experience, to what each technology meant in the bigger picture of a person’s life.

To the people buying and using the “clearly inferior” technology, exactly the opposite was true.

To the user, the interface is the product.


4. Why is there not more Linux on the desktop?

For open source software to take over the world, we’re going to have to do a lot better at user interfaces than we have been doing.

How do I know?

Open source has already taken over the invisible parts of the world: the servers, the infrastructure, the things users need not touch directly.

Mozilla, the most user-experience-focused of open-source companies, has the most adoption by end-users.

People say things to me like, “Linux is only free if the value of my time is zero.”

These are not coincidences.

At one time, the way of open-source software development was thought impossible. But the techniques were invented. The way became possible; then it became successful. Now the techniques are becoming widely known.

The way to make open-source UI design successful is still unclear. We must invent the techniques.


5. Are users dumb?

User interface design is not about dumbing things down for the poor stupid user.

We software developers, understanding the software as we do, find it easy to look down upon those who lack our understanding.

This is wrong.

Users aren’t dumb. They just have better things to do with their lives than memorizing the internal data model of our screwy software.

When software is hard to use, don’t make excuses for it. Improve it.

When a user makes a mistake, don’t blame the user. Ask how the software misled them. Then fix it.

The user’s time is more valuable than ours. Respect it.

Good UI design is humble.


6. Is UI design marketing?

User interface design is not marketing.

Software developers loathe marketing, so if they think that UI design is marketing, then they will loathe UI design.

The qualities of software that make for a good advertisement or computer-store demo are not the same qualities that make software usable and pleasant to work with long-term, day-in day-out. Often these qualities are opposites.

A shopper may choose the microwave with more buttons, because it seems “more powerful”. However, the shopper will soon find out that it does the same thing as any other microwave, you just have to spend longer figuring out which button to push.

It is easy to fool people into buying something that is against their own best interest.

Don’t do that.


7. What is the task of the UI designer?

Let us talk about that microwave some more.

The microwave with the most buttons may be most popular, but it is not the best microwave.

The best microwave has no buttons at all.

It doesn’t need any buttons because it already knows how long you want your food cooked and how hot. You never need to set the clock, either: it’s just always right.

The no-button microwave may not be reachable, but like a guiding star it shows us the direction we should travel.

Users do not know what interface they want. Users do not know what features they want.

Users know the tasks they want to do, and the problems they have.

We learn more by watching the user work than by asking the user.

The job of the UI designer is to provide what the users need, not what the users say they need.

It is to make tasks easier, not to provide features.


8. Where is the science?

User interface design can be approached scientifically. But usually isn’t.

Until we observe people using our software for real, our design is guesswork and superstition.

These things can be measured and given numbers:

  • What program features are being used most frequently, and least.
  • The number of mouse/keyboard interactions required to perform a task.
  • The time it takes a user to figure out how to do a task.
  • Rates of error.
  • How quickly task-completion-time and error-frequency decrease as a user gains experience.

An interface’s efficiency and learnability are empirically determinable quantities.

They are not matters of opinion.

Every user is different, but that’s why we have statistical methods.

The science of design can tell us that interface foo is X% more efficient than interface bar, but bar is Y% more learnable than foo.

Choosing between foo and bar — that’s where the science ends and the art begins.


9. Is change good or bad?

Change has a cost. Change disrupts the user’s habits. Change forces the user to learn something new.

Sometimes the new UI is so much better than the old one that the change is worth the cost.

Sometimes it isn’t.

The trick is knowing when change is worth it.


10. What is the evil of the bad interface?

It is a sin to waste the user’s time, break the user’s train of thought, or lose the user’s work.

Bad user interfaces do all three. Frequently.

Most interfaces are bad.

I do not use the word “sin” lightly.

Because of bad user interfaces, an action taken based on a reasonable assumption or out of habit often results in broken trains of thought, wasted time, and lost work. This is called “user error”, but it isn’t. It is programmer or designer error.

When we blame the user, we teach them that technology is perfect and that the errors are their own. Because technology is hard to use, we are teaching a generation to be afraid of technology. We are teaching a generation to believe in their own stupidity. This is a sin, too.

It’s not the user’s fault.

Tired of Subscription Software?

As Ty mentioned in the January SMUG meeting, Affinity Photo is a very respectable Photoshop replacement. It has, in my opinion, a vastly superior help system with a vast array of tutorial videos linked to the help menu. It even accepts Photoshop plug-ins and I’ve yet to find something I could do in Photoshop that I can’t do in Affinity Photo.

As I write this version 2 of Afinity Photo is launching and until January 25 they have some very attractive pricing. The MacOS license is $40.99, and the iPadOS license is $11.99. This is not a subscription, a one-time fee gives you a perpetual license to the current major version. V1 came out in 2015, and V2 was released last November. So the cost (even if you miss the January 25 deadline) is a small fraction of Adobe’s subscription, especially for occasional users.

As with PS there are sister products, Affinity Designer, an Illustrator-like program, and Affinity Publisher, an Indesign-like program. Until the 25th, a universal license to all three products on three operating systems (MacOS, Windows, and iPadOS) is $99.00.

Regular pricing is $69.99 for the MacOS apps, $19.99 for iPadOS, and $169.99 for the universal license.

The footnotes in the Wikipedia article have links to lots of reviews.

Charging Changes

The European Union has just issued a requirement that electronic devices standardize their charging connectors and they’ve chosen the USB-C connector as the new standard. Because the EU is such a large market for electronic devices this means that manufacturers will be adopting the standard world-wide, including in the US.

This means that laptops, phones and other devices that use external chargers must migrate to USB-C connections over the next couple of years.

The current M1 apple notebooks already use USB-C as their charging connector, but even so, Apple isn’t happy that they will be forced to abandon the current lightning connector used on iPhones and iPads, at least for the European market. It remains to be seen if they will ultimately produce devices with different connectors for the European market than they sell here in the U.S.

Europe already requires all phones to be able to be charged with a USB-micro connector, but there’s a loophole in the rule that enables Apple to comply by including a USB-micro to lightning adapter with each iPhone sold to customers in Europe. The new rule appears written to foreclose that possibility.

EU studies show that EU residents currently dispose of 11,000 metric tons of charging cables each year and the rule is expected to dramatically reduce that number. Margrethe Vestager, the executive vice president for a Digital Europe, said in a press release, “European consumers have been annoyed for long enough by the accumulation of incompatible chargers in their drawers. We have given the industry plenty of time to come up with its own solutions, but now the time has come for legislative action in favor of a universal charger. This is a significant gain for our consumers and our environment, in line with our ecological and digital ambitions.”

But there may be a loophole Apple can slip through in the new rule too. The rule only seems to apply to devices with charging connectors. If Apple switches to wireless charging as used on the Apple Watch then they can build a phone with no connectors at all.

Processor Progression

Apple recently completed the shift from Intel X86 processors to their new family of M1 RISC chips. RISC is an acronym for Reduced Instruction Set Computer.

This is the third time Apple has switched the Macintosh to a new instruction set architecture. The first was from the Motorola 68000 series to PowerPC chips in 1994 and the second from PowerPC to Intel processors using the x86 architecture in 2005–2006.

With the shift from the Motorola 68000 series to the PowerPC, Apple introduced a new Macintosh O/S, System 7 (Later renamed MacOS 7).

On the last day of March this year, Mihai Parparita released two utterly spectacular web-based emulation systems that can run System 7 and it’s successor MacOS 8 in a browser. There’s a demo on YouTube, and you can try them yourself at https://system7.app and https://macos8.app. Each has some pre-installed apps, including Adobe’s Photoshop 3. You can drag and drop files from your desktop to the browser and edit them in the emulation. Uploading and downloading files is supported as well as session-to-session persistent storage.

Read details about the current state of the project, and the work on which it is based in Mihai’s blog.