What’s the place for design in the AI revolution?
Later this week in Sydney I’m MCing From Paint to Pixels: AI & The Creative Industries, a discussion with Jessica Priebe – Art historian and lecturer in emerging technologies at the National Art School, Jamie Sissons – Founder of Absolutely AI, and Jesica Edwards, creative technologist at Canva.
I’ll ask them their thoughts about the professional challenges and opportunities they think generative AI presents creative professionals. It’s free, on Thursday the 27that 6pm, and part of Spark Festival’s AI&U series of events coming up over the next few weeks, so check out the growing list of events, and get along!
So today Ive rounded up some recent articles I’ve come across that relate to the topic of creative professions and AI.
Inside the secret list of websites that make AI like ChatGPT sound smart
Traditional media doesn’t always have a stellar track record at even handed non hysterical approaches to emerging technology, but despite the headline this piece by the Washington Post is thorough, detailed research into
Google’s C4 data set, a massive snapshot of the contents of 15 million websites that have been used to instruct some high-profile English-language AIs, called large language models, including Google’s T5 and Facebook’s LLaMA. (OpenAI does not disclose what datasets it uses to train the models backing its popular chatbot, ChatGPT)
It highlights concern about the western-centric nature of the content, and about racist and other highly problematic content that may have escaped filtering. If we are going to rely on, or even simply use these kinds of tools, it’s our responsibility to have some sense of what is powering them, what biases (and worse) they may have, and be attuned to that.
The Post found that the filters failed to remove some troubling content, including the white supremacist site stormfront.org No. 27,505, the anti-trans site kiwifarms.net No. 378,986, and 4chan.org No. 4,339,889, the anonymous message board known for organizing targeted harassment campaigns against individuals.
You can also determine how much any given site is part of the dataset–I was surprised webdirections.org was ranking at around 400,000 with nearly 60,000 token used from our site. That would be a small subset of the amount we’ve published over nearly 20 years, but also more than I expected.
Making chatbots accessible
Even before the recent round of hype enthusiasm for chat based generative AI, chatbots had become widely deployed as ways for users to interact with services. But how accessible are they? And how can they be made more accessible? While they are text based web content, the kind that in theory should be relatively straightforward to make accessible, they’re interactive, and update content over time, areas which present challenges for ensuring accessibility.
In this piece Camryn Manker looks at challenges for accessibility in building chatbots and ways of addressing these, including various ARIA features.
UX principles for AI products
Right now the focus on generative AIs and large language models is mostly technical, and on what the core technologies can do. While Adobe recently announced Firefly, the generative AI for images currently getting a lot of the attention is Midjourney–which you use by interacting with chatbots in a Discord server.
And the way in which we use almost all these technologies right now is with text prompts.
At the beginning of the personal computer age was the command line prompt. The blinking cursor awaiting the user’s input, which required a knowledge of arcane instructions. While prompt based interfaces don’t require specific instructions, in many ways the product experience is very similar to DOS. We’ve come a longways in terns of UX since then with applications–what will the product experience of AI powered apps look like?
As designers, there are few things more exhilarating than working in new frontiers of technology. What are key UX design principles that will help designers create delightful, highly usable, inclusive experiences for AI?
Product designers, product managers, developers, content professionals, we should all be thinking about this right now.
Is AI driving us away from designing human-centric experiences?
But ironically, asks Dina Gkritzapi, are these technologies driving us away from designing human-centric experiences? She and her colleagues
looked at our design process and defined the way we want to interact with this technology while not allowing it to drive us away from staying human-centric. So here’s our take on how to do that.
They conclude
As digital experience creators, we should not fear these tools but rather embrace them as assistants to our craft. We should strive to understand how these tools work, critique them, address their shortcomings and shape them towards becoming better. With an open and responsible mindset, we can use AI as a tool to assist us in reaching our “North Star”: creating great human-centred design solutions.
We’ll cover all this and much more at Web Direction Summit
We’ve covered AI and machine learning, for designers, developers and product managers for some time now–with two dedicated conferences, and dozens of presentations on Conffab (most available with a free membership).
And we’re making the impact of these technologies on every aspect of web and product design and development a central conversation at Web Directions Summit, returning to Sydney in late October.
Super early bird tickets have just gone on sale–so to make sure you , your teaming organisation stay up with what’s happening, make sure you get along (or stream the event).
Great reading, every weekend.
We round up the best writing about the web and send it your way each Friday.
