7 Trends to Watch in AI & Design for 2024
As we've just wrapped up a frantic year in AI, I wanted to stop and peer into the new year.
What are the biggest trends for 2024 in the Venn diagram of this newsletter - AI, design, and innovation?
I think there are seven major storylines we should pay attention to as we start our next orbit around the sun:
Conversational experiences
Intelligent devices
Race of the frontier models
Blending modalities
Design AI tools
Everyday productivity
Legislation catching up
1. Conversational experiences
What it is
Over the last year, tools like ChatGPT catapulted generative AI into the hands of consumers.
Next, builders of digital experiences will increasingly bake generative AI into the products and services we use.
Conversational Experiences mean we’ll increasingly interact digitally with organizations through an AI-powered conversation rather than a graphical user interface.
The first wave of this movement has been chatbots that are suddenly getting smarter with the help of Large Language Models.
This year, we'll see early adopters embed conversational interactions at the core of digital experiences. Instead of clicking around to find the perfect flight, you’ll tell an AI agent that you’d like to have a family vacation in the sun sometime in November. The agent will find the best flights and hotel recommendations in a natural conversation.
Large Language Models can also improve and augment human communication. A recent example of this is Summer Health, which uses GPT-4 to improve doctors raw notes to be more understandable and cohesive.
The Summer Health app uses GPT-4 to translate doctors' visit notes into more cohesive and understandable patient explanations. Doctors always review the notes to ensure high quality (Source: OpenAI).
Why it matters
The graphical user interface has been the dominant paradigm for getting stuff done online since the Mosaic, the first web browser to show images. It’s hardly the optimal way of achieving our goals.
Generative AI and the powerful array of APIs in developers' hands are poised to transform this. Increasingly, we’ll be able to jump over the abstraction of the website or app to get right to what we want - ordering food, finding a hotel, or understanding the analytics to make business decisions.
Things to consider
Imagine your current digital customer experience; how could it work as a conversation?
How could your customers get value directly from an AI conversation compared to a graphical user interface?
How could you help customers in a more personal and profound way through an intelligent AI conversation?
How can you blend the traditional graphical user interface with conversational elements for the best possible experience?
2. Intelligent devices
What it is
Cheaper sensors and chips brought us smarter devices - known as the Internet of Things. This means many devices, from cars to toothbrushes, are connected to the internet.
Next, more and more devices will have intelligent conversations.
Recently, Microsoft and Google introduced AI models that are small enough to run locally on devices. Innovations in training data curation and model scaling are driving this trend of miniaturization.
This means smaller models can be used on various devices, from smartphones to wearables and beyond.
In the not-so-distant future, your elevator could politely inform you that it has some issues and how to fix them. Devices could communicate in more intricate ways between themselves and humans.
Why it matters
Increasingly, smaller models can perform locally on devices without connection to the internet. Designers of products and digital services can layer intelligent conversations as an interaction mode that can unlock unforeseen opportunities.
Things to consider
If your customer experience includes hardware, consider how even a simple conversational layer could elevate the experience.
In the near future, you can assume most smart devices will have some level of conversational AI capabilities built in. How will it change your customer experience?
3. Race of the frontier models
What it is
The launch of ChatGPT 3.5 just over a year ago started a global arms race in the leading AI models.
OpenAI is still in pole position with its GPT-4 model, with Google’s announced but not yet launched Gemini Ultra on par or a few steps behind.
In 2024, we can expect major advances and launches from OpenAI. Whether it’s a significant leap like GPT-5 or a more incremental improvement remains to be seen.
What is clear is that we’re bound to see years of bloody battles between the giants in the field.
On the high end, models will continue to grow more capable, and on the other low end, we’ll see cheaper and smaller models improve.
Things to consider
If you’re trying to sell your organization on the need to invest in AI and design capabilities, it can be useful to paint a picture of how fast the field is progressing
There is a plethora of models to build on top of now. It makes sense to consider what level of capability makes sense for your task at hand
Why it matters
If you’re designing services or strategies now, it’s safe to assume the AI models we’ll have as building blocks are only getting smarter.
As useful as AI tools like LLM’s are now, you can expect them to be profoundly more impactful with the next iterations of the models.
4. Blending modalities
What it is
As we discussed in the newsletter about Google’s recent Gemini model, Large Language Models are increasingly morphing into all-encompassing everything models.
This means they can create text, images, sound, and code. They can also take these different modalities as inputs.
Why it matters
If you’re designing digital experiences, being able to pull from these different modalities can unlock interesting possibilities. Customers could, for example, upload a picture of their clothing and get tips on complementing their style with new purchases.
For the design work itself, you can increasingly feed AIs your design in textual and visual form to elicit feedback. The ability of AI models like ChatGPT and Gemini to produce visual designs of, say, UIs is still quite rudimentary - though expected to improve.
Things to consider
The multimodal capabilities of AI models are brand new. Next year, we should experiment with how we can use these newfound skills to improve our work and our customers' experiences.
5. Design Specific AI tools
What it is
Currently, the most useful AI tool in most designers' toolkits is ChatGPT or a general image generation tool like Midjourney.
We’ve seen some early attempts at design-specific tools like Relume Library, but nothing truly transformational has popped up.
In 2024, I’m expecting at least Figma to ship something significant in the realm of product design. They already teased the UI design equivalent of Photoshop generative fill.
The design of user interfaces has multiple layers of complexity compared to many other forms of design. Still, I’m absolutely convinced AI will be able to aid UI in design, starting with simple UI pattern generation and progressing into more complex flow designs.
Why it matters
In 2024, most people working in design will have a chance to try powerful design-specific AI tools. They’ll also have a chance to reflect on their own future role in the design of products and services.
Things to consider
A few years into the future, I expect AIs to handle most of our individual hands-on design tasks. We’re freed up to orchestrate the bigger picture, workshop with stakeholders, empathize with users and steer the AIs in the right direction.
6. Everyday Productivity
What it is
Next year, AI will be embedded deeper into the everyday work of people working on design and innovation.
2023 was the year of excitement and experimentation with using generative AI in our workflows. In 2024, we’ll go deeper into systematically adopting AI as a pivotal tool in our work. Organizations will also invest more in ramping everyone up on using AI - not just the people who follow newsletters like these.
Last year, companies already saw dramatic productivity increases in some areas. For example, Goldman Sachs reported a 20-40 % increase in developer productivity with the help of generative AI.
Why it matters
This year, all of us working in design and innovation will need to become increasingly comfortable working alongside AIs. We’ll train our helper agents and develop a stronger intuition of when and how to employ AIs.
Things to consider
Think about how you could, as an early adopter, help your organization scale your maturity in AI on different levels, from everyday productivity to powering better customer experiences.
7. Legislation catching up
What it is
AI is a slippery soap to legislate. It’s so powerful and profound it needs some sensible legal guardrails - akin to medicine and flying. But how do you legislate something moving at such a highly complex and unpredictable pace?
The EU is taking a first meaningful stab at this. The EU AI act is a powerful and far-reaching legislative agreement that will likely soon be voted into law in the EU.
Among other things, it attempts to ban potentially harmful uses of AI, from mass surveillance to social scoring.
High-risk uses of AI will be under closer legal scrutiny and must follow certain obligations to ensure their safety. Also, general-purpose AI systems like GPT from OpenAI will have to adhere to legislation around things like transparency, testing, and copyright.
How exactly the EU AI act will play out remains to be seen.
Why it matters
Anyone involved in using AI at any meaningful scale and public capacity must pay attention to the EU AI Act. It will shape the guardrails of AI use within the EU and potentially worldwide.
Things to consider
Follow this newsletter and other outlets to follow the impact of the EU AI act on product and service innovation. Pay particular attention if you’re working in an area that might be categorized as high-risk.
Last year, I worked with the AI governance company Saidot - an excellent partner for any organization that wants to ensure compliance with the shifting regulations.
The year ahead
2023 was a monumental year in AI. Still, we’ve probably just seen the early rumblings of earth-shattering movements.
Expect companies to adopt AI in a more profound way this year. The models will become smaller, more powerful, efficient, capable, and well-versed in all media types.
Still, it’s all just getting started. We haven’t even scratched the surface of opportunities for designing better experiences with AI.
A massive thank you for following the world of AI and design with me this year, and have a great start to 2024!
Want to make sure you catch my next article?
This article was first published on my weekly newsletter on how to use generative AI for design and innovation.
