Statement on Generative AI | Ben Myers
I endorse this statement.
I endorse this statement.
It looks like it will be a great tool for prototyping. A tool to help developers that don’t have experience with CSS and layout to have a starting point. As someone who spent some time building smoke and mirrors prototypes for UX research, I welcome tools like this.
What concerns me is the assertion that this is production-grade code when it simply is not.
The slides and transcript from a great talk by Maggie Appleton, including this perfect description of the vibes we get from large language models:
It feels like they’re either geniuses playing dumb or dumb machines playing genius, but we don’t know which.
Another great talk from Simon that explains large language models in a hype-free way.
I just described prototype code as code to be thrown away. On that topic…
I’ve been observing how people are programming with large language models and I’ve seen a few trends.
The first thing that just about everyone agrees on is that the code produced by a generative tool is not fit for public consumption. At least not straight away. It definitely needs to be checked and tested. If you enjoy debugging and doing code reviews, this might be right up your street.
The other option is to not use these tools for production code at all. Instead use them for throwaway code. That could be prototyping. But it could also be the code for those annoying admin tasks that you don’t do very often.
Take content migration. Say you need to grab a data dump, do some operations on the data to transform it in some way, and then pipe the results into a new content management system.
That’s almost certainly something you’d want to automate with bespoke code. Once the content migration is done, the code can be thrown away.
Read Matt’s account of coding up his Braggoscope. The code needed to spider a thousand web pages, extract data from those pages, find similarities, and output the newly-structured data in a different format.
I’ve noticed that these are just the kind of tasks that large language models are pretty good at. In effect you’re training the tool on your own very specific data and getting it to do your drudge work for you.
To me, it feels right that the usefulness happens on your own machine. You don’t put the machine-generated code in front of other humans.
Emily M. Bender:
I dislike the term because “artificial intelligence” suggests that there’s more going on than there is, that these things are autonomous thinking entities rather than tools and simply kinds of automation. If we focus on them as autonomous thinking entities or we spin out that fantasy, it is easier to lose track of the people in the picture, both the people who should be accountable for what the systems are doing and the people whose labor and data are being exploited to create them in the first place.
Alternative terms:
And this is worth shouting from the rooftops:
The threat is not the generative “AI” itself. It’s the way that management might choose to use it.
This is a really clear, practical, level-headed explanatory talk from Simon. You can read the transcript or watch the video.
Could the tsunami of AI shite turn out to be a flash flood? Might the models rapidly degrade into uselessness or soon be sued or blocked out of existence? Will users rebel as their experience of the internet is degraded?
In my most optimistic moments, I find myself hoping that the whole AI edifice will come tumbling down as tools disintegrate, people realise how unreliable they are, and how valuable human-generated and curated information really is. But it’s not a safe bet.
Chat is rarely a suitable interface for most tasks. Here, Maggie Appleton explores and prototypes some alternatives.
AI is great anything quantity-related and bad and anything quality-related.
Sensible thinking from Dan here, that mirrors what we’re thinking at Clearleft.
In other words, it leans heavily on averages; the closer the training data matches an average, the higher degree of confidence that the result is more “correct,” or at least desirable.
The problem is that this is the polar opposite of what we consider creativity to be. Creativity isn’t about averages. It’s about the outliers, sometimes the one thing that’s different than all the rest.
Generative AI: What You Need To Know is a free resource that will help you develop an AI-bullshit detector.
You can read all the cards on one page, print them out, or print to PDF.
I’m sure you’ve heard the law of the instrument: when all you have is a hammer, everything looks like a nail.
There’s another side to it. If you’re selling hammers, you’ll depict a world full of nails.
Recent hammers include cryptobollocks and virtual reality. It wasn’t enough for blockchains and the metaverse to be potentially useful for some situations; they staked their reputations on being utterly transformative, disrupting absolutely every facet of life.
This kind of hype is a terrible strategy in the long-term. But if you can convince enough people in the short term, you can make a killing on the stock market. In truth, the technology itself is superfluous. It’s the hype that matters. And if the hype is over-inflated enough, you can even get your critics to do your work for you, broadcasting their fears about these supposedly world-changing technologies.
You’d think we’d learn. If an industry cries wolf enough times, surely we’d become less trusting of extraordinary claims. But the tech industry continues to cry wolf—or rather, “hammer!”—at regular intervals.
The latest hammer is machine learning, usually—incorrectly—referred to as Artificial Intelligence. What makes this hype cycle particularly infuriating is that there are genuine use cases. There are some nails for this hammer. They’re just not as plentiful as the breathless hype—both positive and negative—would have you believe.
When I was hosting the DiBi conference last week, there was a little section on generative “AI” tools. Matt Garbutt covered the visual side, demoing tools like Midjourney. Scott Salisbury covered the text side, showing how you can generate code. Afterwards we had a panel discussion.
During the panel I asked some fairly straightforward questions that nobody could answer. Who owns the input (the data used by these generative tools)? Who owns the output?
On the whole, it stayed quite grounded and mercifully free of hyperbole. Both speakers were treating the current crop of technologies as tools. Everyone agreed we were on the hype cycle, probably the peak of inflated expectations, looking forward to reaching the plateau of productivity.
Scott explicitly warned people off using generative tools for production code. His advice was to stick to side projects for now.
Matt took a closer look at where these tools could fit into your day-to-day design work. Mostly it was pretty sensible, except when he suggested that there could be any merit to using these tools as a replacement for user testing. That’s a terrible idea. A classic hammer/nail mismatch.
I think I moderated the panel reasonably well, but I have one regret. I wish I had first read Baldur Bjarnason’s new book, The Intelligence Illusion. I started reading it on the train journey back from Edinburgh but it would have been perfect for the panel.
The Intelligence Illusion is very level-headed. It is neither pro- nor anti-AI. Instead it takes a pragmatic look at both the benefits and the risks of using these tools in your business.
It has excellent advice for spotting genuine nails. For example:
Generative AI has impressive capabilities for converting and modifying seemingly unstructured data, such as prose, images, and audio. Using these tools for this purpose has less copyright risk, fewer legal risks, and is less error prone than using it to generate original output.
Think about transcripts of videos or podcasts—an excellent use of this technology. As Baldur puts it:
The safest and, probably, the most productive way to use generative AI is to not use it as generative AI. Instead, use it to explain, convert, or modify.
He also says:
Prefer internal tools over externally-facing chatbots.
That chimes with what I’ve been seeing. The most interesting uses of this technology that I’ve seen involve a constrained dataset. Like the way Luke trained a language model on his own content to create a useful chat interface.
Anyway, The Intelligence Illusion is full of practical down-to-earth advice based on plenty of research backed up with copious citations. I’m only halfway through it and it’s already helped me separate the hype from the reality.
In some ways, the fervor around AI is reminiscent of blockchain hype, which has steadily cooled since its 2021 peak. In almost all cases, blockchain technology serves no purpose but to make software slower, more difficult to fix, and a bigger target for scammers. AI isn’t nearly as frivolous—it has several novel use cases—but many are rightly wary of the resemblance. And there are concerns to be had; AI bears the deceptive appearance of a free lunch and, predictably, has non-obvious downsides that some founders and VCs will insist on learning the hard way.
This is a good level-headed overview of how generative language model tools work.
If something can be reduced to patterns, however elaborate they may be, AI can probably mimic it. That’s what AI does. That’s the whole story.
There’s very practical advice on deciding where and when these tools make sense:
The sweet spot for AI is a context where its choices are limited, transparent, and safe. We should be giving it an API, not an output box.
Maggie Appleton:
An exploration of the problems and possible futures of flooding the web with generative AI content.
Writing, both code and prose, for me, is both an end product and an end in itself. I don’t want to automate away the things that give me joy.
And that is something that I’m more and more aware of as I get older – sources of joy. It’s good to diversify them, to keep track of them, because it’s way too easy to run out. Or to end up with just one, and then lose it.
The thing about luddites is that they make good punchlines, but they were all people.
I feel like there’s a connection here between what Kevin Kelly is describing and what I wrote about guessing (though I think he might be conflating consciousness with intelligence).
This, by the way, is also true of immersive “virtual reality” environments. Instead of trying to accurately recreate real-world places like meeting rooms, we should be leaning into the hallucinatory power of a technology that can generate dream-like situations where the pleasure comes from relinquishing control.
Baldur has new book coming out:
The Intelligence Illusion is an exhaustively researched guide to the business risks of Generative AI.
I’ve mentioned before that I’m not a fan of initialisms and acronyms. They can be exclusionary.
It bothers me doubly when everyone is talking about AI.
First of all, the term is so vague as to be meaningless. Sometimes—though rarely—AI refers to general artificial intelligence. Sometimes AI refers to machine learning. Sometimes AI refers to large language models. Sometimes AI refers to a series of if
/else
statements. That’s quite a spectrum of meaning.
Secondly, there’s the assumption that everyone understands the abbreviation. I guess that’s generally a safe assumption, but sometimes AI could refer to something other than artificial intelligence.
In countries with plenty of pastoral agriculture, if someone works in AI, it usually means they’re going from farm to farm either extracting or injecting animal semen. AI stands for artificial insemination.
I think that abbreviation might work better for the kind of things currently described as using AI.
We were discussing this hot topic at work recently. Is AI coming for our jobs? The consensus was maybe, but only the parts of our jobs that we’re more than happy to have automated. Like summarising some some findings. Or perhaps as a kind of lorem ipsum generator. Or for just getting the ball rolling with a design direction. As Terence puts it:
Midjourney is great for a first draft. If, like me, you struggle to give shape to your ideas then it is nothing short of magic. It gets you through the first 90% of the hard work. It’s then up to you to refine things.
That’s pretty much the conclusion we came to in our discussion at Clearleft. There’s no way that we’d use this technology to generate outputs for clients, but we certainly might use it to generate inputs. It’s like how we’d do a quick round of sketching to get a bunch of different ideas out into the open. Terence is spot on when he says:
Midjourney lets me quickly be wrong in an interesting direction.
To put it another way, using a large language model could be a way of artificially injecting some seeds of ideas. Artificial insemination.
So now when I hear people talk about using AI to create images or articles, I don’t get frustrated. Instead I think, “Using artificial insemination to create images or articles? Yes, that sounds about right.”
This is a genuinely lovely use of machine learning models: provide a prompt for an illustration to print out and colour in.
Mike explains his motivation for building this:
My son’s super into colouring at the moment and I’ve been struggling to find new stuff for him.
In which Rob takes a deep dive into isometric projection and then gets generative with it.