R/T 008 - The AI Takeover of Design
This week's radical thing is: Exploring the potential impacts of Artificial Intelligence on the design industry
Everyone is losing their marbles over artificial intelligence (AI).
Many of you will have become familiar with generative AI through the tidal wave of interest created by Chat GPT. Or you might be exploring image creation with DALL-E 2, Stable Diffusion or Midjourney.
Every LinkedIn influencer and their dog are trying to stake their claim to predict the future - which is a fool's errand.
Yes there is a lot of hype. There’s also a wide variety of expert opinion. Yet compared to other hyped technologies, such as Blockchain, Crypto and NFTs, the current AI wave appears less opaque, more accessible and presents numerous use cases.
There are many factors complex factors related to the development of AI tools. These include the legality of the creation of AI training data, the ethics of using tools built on these data sets and the copyright of works created. Each topic warrants examination, but I want to focus on one area specifically.
CBI’s breakdown of Y-Combinator’s AI investments
CBI recently published a visualisation of Y-Combinator's generative AI investment thesis. It shows where the start-up incubator places its bets on which industries AI will disrupt first.
'Engineering, Product and Design" makes up almost 40% of the folio.
In this edition of Radical Things, I'm examining three things I've noticed recently in AI and unpacking their implications for design.
Microsoft Designer
I regularly pour scorn on Microsoft Teams, but I'm currently part of the beta launch for Microsoft Designer.
Microsoft Designer feels like a combination of Photoshop, InDesign and Keynote. You provide a chat prompt, and it produces a page layout for you. It then creates a series of AI-driven alternatives you can choose and build on. A DALL-E 2 image engine creates your custom images.
As you can see from the image below, it looks nothing like a Microsoft product from the promotional material. It's much more contemporary and design-led. It feels like something executed by Apple rather than Microsoft.
Take particular notice of the value proposition:
"Create stunning designs in a flash with Microsoft Designer
No design skills needed! And if you have design skills, we're here to help you flex 'em."
So the programme saves you time. You don't need any skills to create stunning designs; if you have design experience, we can augment and improve you.
A promotional splash page for Microsoft Designer
Auto GPT
In the last week, Auto GPT has begun trending across Twitter.
One of the (many) complaints about Chat GPT is its inability to 'remember' previous inputs, as no memory is assigned to it—this lack of capability results in constantly repeating prompts where you receive different answers.
Auto GPT is a self-prompting version of Chat GPT that also contains memory and has access to the internet. It's an AI Agent, which might be a stepping stone towards AGI (Artificial General Intelligence).
Here are some examples:
A to-do list that does itself.
A market researcher that anticipates bias and solves its own problems along the way.
Build a custom, personalised website.
Image: Levi Strauss & Co. / Lalaland.ai
Levi’s AI Models
The denim dominator recently announced their partnership with LaLaLand.ai - a digital fashion studio that creates AI-generated models.
Although Levi's has stated they will only use these synthetic models to 'supplement human models', many people are left wondering, WTF? The research, efforts and resources in creating these AI models could have gone into, guess what?
Hiring diverse human models!
But maybe humans aren’t so sustainable, which is part of the pitch for the use case, as well as the goal of ‘increasing diversity’. If the images we see are no longer real, is this goal even achievable?
So, why should we care about any of this?
Firstly, we're facing rapid development and implementation of AI models across different industries. Last year I was running a Twitter space discussing the impacts of AI on the design industry. For those few months, progress was rapid - with Chat GPT4, new developments are accelerating.
Even in the example of AutoGPT, we can see the earliest nascent form of AGI, looking after itself, creating tasks, solving problems and addressing new ones as they emerge.
Maybe these are the 'Muppet Babies' of AI?
In the Microsoft example, we see the removal of expertise as a barrier to designing. One of the biggest enterprise software companies on earth is the delivery channel for this. Everyone with 'Word' on their computers may soon have access to this technology.
Finally, we have humans supplemented for representations of humans. By removing humans from the modelling value chain, we remove their agency and potential for success and career building.
Overlap these three things, remove 'design' as a placeholder, and apply them to other industries and sectors.
You start to get the picture.
If we return to design for a moment - some of the implications for the industry might look like this:
The rapid evolution of AI accelerates the commodotisation of specific design skills.
Low end-design is executed exclusively by AI and low-skilled operatives (Microsoft Designer).
Design agencies engage in deep design specialisation augmented by AI and remain relatively insulated from significant disruption.
Pressure from the top and bottom creates a "hollowing-out of the middle."
AI will likely only partially replace humans in the various disciplines that make up the industry. Design is a very human capability. It’s also relational and requires the navigation of complicated and complex problem domains.
That said, specific skills and roles will grown more susceptible to what is already happening. There’s also no saying that with enough computational power we couldn’t brute force creativity, which would create it’s own existential crisis.
Education and training will be critical in preparing designers for the industry's changing nature, especially as AI models and tools become more powerful and tightly integrated into a designer's workflow. But how fast can educational institutions move and anticipate these changes, and what might fill the void?
Finishing this edition is exceptionally challenging. Every time I write a closing line, I realise the vast assumptions that sit behind it:
There will always be a need for specialisation.
Companies will always require the discipline of design.
Design will be amortised into a more general problem-solving approach.
And there lays the challenge.
There's so much rapid change, continuing uncertainty and all our assumptions are open to challenge.
Collective thinking, discourse and action will help us navigate the road ahead, but for the moment - the future seems up for grabs for anyone in possession of an LLM.
What do you think about the implications of AI for your job, sector or industry?
Drop your thoughts in the comments - I’d love to hear them.