AIoT
Design for the Artificial Intelligence of Things
This week’s issue of Design Things is brought to you by my friends at Moo’ed.ai.
Moodboards are a fundamental tool for creatives, but in today's fast-paced world, the process of sifting through a vast sea of inspiration can be time-consuming and daunting. Mooed.ai is a revolutionary solution that empowers users to effortlessly create mood boards with just a few clicks. Not only does it save valuable time, but it also proves to be cost-effective by enabling high-volume output, breaking free from creative constraints. The magic lies in the meticulously crafted user interface, born out of the minds of fellow creatives. It is designed by designers, for designers.
If you were a casual observer looking at the AI hype-sphere today, you might be forgiven in thinking that it is largely a software phenomenon.
The big story is that breakthroughs in AI technology (mainly advances in deep learning, a rapid increase in computing power that allowed processing of huge datasets, and widespread access to LLMs) has caused a boom in the development of AI-powered software and a complete paradigm shift in how we think about interacting with that software.
If you are a digital designer, you might be rejoicing in this new development. Big companies like Microsoft and Google investing heavily into AI, more investment money flowing into the space, more new startups, more opportunities for you.
For awhile now we’ve seen the opportunities and rewards flow steadily towards towards the digital side of design. The recent AI boom might seem like an extreme acceleration of that trend.
If you are an industrial designer or hardware engineer, you might be panicking. Another nail in the coffin of designing for the physical world. Another year of watching the party from the outside.
Time to sign up for that UX bootcamp?
No, not yet.
AIoT is coming
In the 2010s, there was a boom in tech hardware investment and development. The buzzword then was IoT, or Internet of Things.
Rapid advances and declining costs in wireless technology, sensor technology, and cloud computing meant that almost every electrical device could and would eventually be connected to the internet.
Smartphone proliferation and improved internet infrastructure allowed almost everyone to have a connected supercomputer in their pockets, and it allowed the Internet of Things to outsource some of its computing and user interface needs to a centralized, portable, and personal device rather than having to build it into the IoT hardware itself.
IoT startups popped up all over the place, inspired by success stories like Nest (acquired by Google in 2014 for a whopping $3.2B), which redefined an unloved product category and sparked a smart home revolution as people started to see their previously boring home appliances as potentially internet-connected, app-controlled, Siri (or Google Home) integrated gadgets.
This was a boom time for anyone involved in hardware.
Every electronic device was being re-imagined as a connected one. New chips and sensors needed to be integrated, buttons, screens, and user interfaces needed to be defined, and everything needed to look new and differentiated from their old, boring, non-IoT versions.
Today in 2023, all this seems commonplace. IoT stopped being exciting. It became table stakes.
Of course you can tweet from your fridge.
Of course your car can be unlocked from your iPad.
Of course you can shoot treats at your cats out of your cat treat launcher at home via your smartphone while you’re halfway across the world on vacation (no joke, seen a friend do it).
But now, a new horizon is coming into sight: AIoT. Artificial Intelligence of Things (which, sadly is not an acronym I came up with).
With new advances in AI technology, I think we are going to see another hardware boom because once again, the underlying hardware needs of the technology have changed and completely new paradigms around how we interact with the hardware are possible.
New capabilities for intelligent hardware
AI promises to bring a dearth of new exciting capabilities to hardware, the most exciting of which to me are intelligent adaptability and more natural/human interactions.
For all the talk about “smart” devices, the IoT revolution was actually pretty dumb. A lot of “smart” devices could collect data via sensors and user input, but their decision making capabilities were limited to pretty simple if-this-then-that kind of logic. There wasn’t really much intelligence there. Tech critics often noted that the smart home wasn’t smart, just connected.
AI-enabled hardware could potentially process a ton more data more efficiently, especially messy, unstructured, real-world data that doesn’t always follow neat patterns. It could then intelligently make decisions about what to do with that information without input from the user.
When the user does need to interact with the device, the potential for more natural interaction patterns is huge.
IoT hardware allowed user interactions, but usually only in a way that forced the user to conform to the input it could receive.
That means you have to consciously think about how to interact with every smart device you own, which adds a ton of cognitive load when everything around you is smart but in different ways.
With AI-enabled hardware, you could potentially interact with everything through natural written or spoken language or even gestures, removing a layer of how to interact with the damn thing.
ChatGPT is already a great example of changing the way many people think about interacting with computers. If you want to learn something, you can actually have a conversation with it, as opposed to Google where you have to formulate a series of queries and judge the relevance of the results.
A relevant hardware example of this is of course, the Humane AI Pin, an OpenAI powered wearable that functions as a smart assistant that you can interact with via voice control and hand gestures.
I haven’t tried it myself, but it seems to me the idea behind this device is to strip away the annoyingly rigid technology layer from our interactions with the digital world and make it more of a seamless, intelligent experience.
This kind of concept isn’t new, but with AI-enabled hardware, it’s actually becoming possible.
Design opportunities for AIoT
But wait. Why do we need new hardware? Isn’t all that AI stuff just being done in the cloud? I don’t need new a new laptop to run ChatGPT.
Part of the reason is fairly pragmatic. In order to process the large amounts of data to make something artificially intelligent and to keep latency low enough to deliver a compelling user experience, a lot of the computation may need to be done locally on-device or at least near it. This also has benefits for privacy reasons, as all that granular data about your interactions aren’t being sent to a centralized location.
This means that current hardware with chips not designed for AI will just not cut it for the AIoT revolution. You can’t just upgrade the firmware and have it now be AI-enabled.
With IoT, you needed all new hardware to add internet connectivity, buttons, screens, LED indicators etc. to the smart device. With AI-enabled hardware, you might need all-new hardware with less user interface but more internal processing power to deliver a new type of experience. This could mean a new generation of hardware that is more ambient and less overtly techy.
A great example of this is the Rewind Pendant, an AI-powered wearable recording device that is supposedly going to help you with recalling and synthesizing important information.
Though I’m not sure it’s anything more than a rendering right now, it is interesting that they’ve gone with such a nondescript aesthetic that doesn’t really look all that techy. I think it represents a more seamless integration of technology and it’s designed to blend more into the background rather than scream, “look at me, I’m a smart device!”.
It makes sense that software and digital are enjoying the fruits of the AI boom right now. Software will generally lead in new technological developments and investment because its simply faster to innovate, iterate, and scale. But a lot of the value of new technological developments is in how they translate to the physical world, where we still very much live. There’s a reason that 50% of the world’s most valuable companies are hardware companies.
With this impending AIoT revolution, a whole new generation of hardware to be designed, and a completely new way of interacting with our hardware gadgets to be defined, I think there is going to be another boom in opportunities for hardware designers and engineers. Especially those that have an understanding of the benefits of AI and its integration with the physical world.
So put away your credit card and close the UX bootcamp signup tab.
There’s work to be done.
3 ways I can help you
LinkedIn for Designers Course My 1-hour downloadable video course that will teach you how to use LinkedIn to grow your design career or business through creating engaging design content.
Book a 1-hour consulting session. Dedicated time to talk about your product, design strategy, business strategy, career or anything else you'd like. Use promo code STUDENT for 80% off if you are a student or young professional (<5 years career experience).
Sponsor this newsletter. Promote your product, service, or offer to 2,500+ designers that get Design Things in their inbox every week. Note: I only promote products that I have personally tested and would recommend.
What did you think of this email?
Your feedback helps me make this newsletter relevant and interesting to you.
Definitely a proponent of more beautifully designed and well performing hardware products but I'm struggling to find what genuine problems AI related hardware is solving. Take Humane for example - it's beautiful, great materials choice, apple like contours but how many people really need a device for the use cases presented.