Some thoughts on AI wearables
For the past couple of months, we’ve started to see a new wave of AI hardware products released into the world. Meta’s new smart glasses, the humane AI pin, and better writers than I have already commented on these things, but as always, I have thoughts.
First, I believe that AI wearables have an actual market. I think both the Meta / Humane products are trending towards this idea of an invisible computer that eliminates the need for a phone in your pocket all the time, and this is something I’d personally really enjoy. I find myself really wanting both of these products, although I will probably only purchase the Meta glasses, if only just to get a pair of transitional lenses with Bluetooth headphones built in. Although the Humane product is closer to what I’d want in terms of functionality, it’s explicitly designed as a complete phone replacement. It uses a separate cellular subscription without number sharing like the Palm phone / Apple Watch, and has no phone app at all to speak of. I can see myself wanting a version of the AI pin maybe half a decade down the line when they eventually figure out that trying to compete against the iPhone is a bad idea. Either that, or the whole thing goes under. I do think Humane might be onto something in terms of dedicated form factor for maybe, 20 years down the line, but this looks like a classic case of thinking the consumer base is more ready for this than they actually are. You need to let them get their feet wet with this idea first, as Meta is, without forcing them to take the plunge. Otherwise, change is scary and the iPhone has Marvel Snap on it, so who are you really gonna convince here.
There’s also the rumors of the OpenAI / Ive device, which right now seems to be a bit of a bust. OpenAI (disclosure: my employer Microsoft has invested in / works with OpenAI, these views do not represent my employer or OpenAI, etc.) has been making the brilliant pivot to a consumer company, and as Casey Newton brilliantly mentioned on this week’s episode of Hard Fork, is turning into a generationally important company. The next Facebook / Google-sized company is sort of forming right in front of us, and it’s incredible to witness as a conscious adult who is now a tech worker. Sam Altman has commented on these rumors saying they just don’t have a good idea for one yet, which doesn’t give me much faith any of these companies do either. Right now, the Meta glasses / Humane AI Pin both seem to be existing hardware ideas that had AI forced into their marketing pitch within the past year. The predecessor to the Meta glasses, the RayBan’s stories, had no such integration, and the Humane AI Pin was pitched as essentially a wearable camera in leaked decks going back to 2020. The AI boom has had the teams of both companies utilizing AI, regardless of how it enhances the product experience, as a way to generate excitement both internally and externally for these new devices rather than it being something consumers truly want.
For the past few months, I have been paying for ChatGPT Plus. I found that it, along with Raycast AI, are the two LLM services that have stuck with me the most. Raycast’s AI features are just convenient as all hell, and they’re exactly where I need them when working, whether it’s to recall a bash command or to formalize my emails. ChatGPT Plus, meanwhile, has turned into a Swiss army knife of sorts, a really good value for the 20 dollars given that it has advanced data analysis (writes python code for you to crunch stats from spreadsheets), browsing (Bing chat but way nicer), a very nice iOS app, an AI voice assistant on mobile that sounds a lot like Scarlett Johansson in Her, unlimited uses of DALL-E 3 for AI image generation, and unlimited GPT-4 uses. The new GPT’s feature is a neat little preview into where they want the future to head, but I haven’t found it to be that genuinely useful yet in my experiments, other than as a way to save and share custom prompts. I’ll hold my judgment until I see what others can do with the platform though. Regardless, I found that the voice model has been really useful, especially with the added browsing. It’s become the thing I use my action button on my iPhone for the most, as unlike Siri, it will actually respond to my information queries with voice rather than giving me a list of search results and telling me to figure it out myself.
In summary, I think AI works alright on my iPhone and my Mac! So why are a bunch of companies working to build these products into wearable devices? First off, I think the appeal partly lies in its speed. Second, it’s Apple. By launching “app stores”, building their own hardware that works independently of the iPhone, these companies are attempting to emulate the success of the iPhone for what they believe to be a new computing paradigm. I don’t think they’re completely wrong for this, but I want to stress something to every single person working on AI products and in the field.
ChatGPT is a 100% free iPhone app.
GPT-4 wrappers are considered by many to be a derogatory term, but I believe it’s an accurate one. So many companies believe they can essentially white label OpenAI with a software app or tool, and ask users to pay even more than ChatGPT Plus for it. You can’t, because ChatGPT is a 100% free iPhone app. Your AI tool in your notebook app you make people pay for forgets this. Your online SQL query writer forgets this. Moreover, AI wearables that compete in pricing with the iPhone seem to overlook the fact that ChatGPT is a 100% free iPhone app. White labeling doesn’t stop being white labeling because it’s a pin on your chest. You can’t compete with a free app that exists on the device people already have and love, unless you increase the convenience factor by an order of magnitude.
All this to say, I think eventually these products have a market someday. I couldn’t even begin to guess when that day will be though. Could be as soon as next year, or as far as 20 years from now. All I can say for sure, is that it’s not now.