Why Is the Apple Vision Pro Going to Get Its Shot, While the Humane AI Pin Is Not Going to?

How much rationality is there in the market reactions so far—rationality in individual assessments and intentions, & rationality in the workings of humanity considered as an anthology intelligence here?…

Subscribe now

Share

I betcha the Humane AI Pin people looked at Apple VisionPro’s pulling in more than $1,000,000,000 in sales revenue in less than two and a half months for something that is not a product, but rather just a kludgy simulation of a product.

I betcha they then said: We can charge $699 plus $24/month for our kludgy simulation of our product too! The same kinds of people who plunk down money to be AVP alpha-testers so that they can pretend they are living in the future will plunk down money to be HAIP alpha-testers too!

Back up:

Between the start of February and mid-April 2024 Apple Computer has managed to sell somewhere north of 250,000 $4,000 Apple VisionPro headsets—for a total revenue of more than one billion dollars, and at least a 300 million-dollar cash-flow offset to the cost of developing the device in the first place. And yet the Apple VisionPro is not the product. The product is an AR glasses-iike “spatial” or perhaps “ambient” computing system that we do not yet know how to make. A future glasses form-factor product that accomplishes the same ambient or spatial computing tasks that the Apple VisionPro accomplishes—plus bonus superior-quality immersive VR features—would indeed be insanely great. But at the moment all we have a simulation of the product, albeit a simulation that can be very nice for those whose eyes and brain can adapt to the experience, who love the immersive VR, who find it a very effective tool for virtually carrying very large external displays around with their laptop, and who don’t mind having two iPads and an immense pair of ski goggles on their face.

But still, $4000.

That brings us to the Humane AI pin:

David Pierce: Humane AI Pin review: not even close: ‘For $699 and $24 a month…. I stand in front of a store or restaurant, press and hold on the touchpad, and say, “Look at this restaurant and tell me if it has good reviews.” The AI Pin snaps a photo… pings… image recognition models, figures [it] out… scours the web… returns[:]… Tacombi has great reviews, it might say. People really like the tacos and the friendly staff…. Very neat… would take much longer and many more steps on a smartphone…. I find I want what Humane is selling even more than I expected. A one-tap way to say, “Text Anna and tell her I’ll be home in a half-hour,” or “Remember to call Mike tomorrow afternoon,” or “Take a picture of this and add it to my shopping list” would be amazing. I hadn’t realized how much of my phone usage consists of these one-step things, all of which would be easier and faster without the friction and distraction of my phone. But the AI Pin doesn’t work. I don’t know how else to say it… <https://www.theverge.com/24126502/humane-ai-pin-review>

Share Brad DeLong’s Grasping Reality

And:

Marques Brownlee: MKBHDs For Everything: ‘The Humane AI Pin… is a brand new product in a really interesting new form factor…. Unfortunately it’s also the new worst product I think I’ve ever reviewed in its current state… <https://stratechery.com/2024/mkbhds-for-everything/>

Leave a comment

Cf. to Marques Brownlee on the Apple VisionPro two months earlier:

Marques Brownlee: ‘Attempting tomorrow’s ideas… with today’s technology This is my thoughts on the weight/straps FYI…. First time I tried Vision Pro: Damn this display is amazing and the eye tracking is like magic and this feels very future and also it’s a little heavy Second time: immersion factor is still so high. Special videos are hit or miss, you gotta get the distance right. And wow this thing is really, heavy, not sure how long I’d be able to wear this Third time: Damn this thing is heavy. Also the typing experience is decent. There’s some new cool apps to check out. But wow. So heavy… <https://twitter.com/search?q=%22vision%20pro%22%20(from%3AMKHBD)>

Refer a friend

I do not think Marques Brownlee is or is going to soon become a frequent AVP viewer. Yet there is still a huge contrast between his view of AVP and his view of HAIP. Perhaps it is as John Gruber said:

John Gruber: Underpromise & Overdeliver: ‘The sort of 1.0 review you want to see: It’s good at what it already does and I can see how it could do more in the future. The one and only review of the Humane AI Pin that expresses a sentiment like that is Raymond Wong’s for Inverse… <https://daringfireball.net/linked/2024/04/14/humane-underpromise-overdeliver>

Give a gift subscription

Perhaps that is the reason for the different reactions. Perhaps to be successful a version 1.0 release that is really a demo of a product in the form of a kludgy simulation of a product has to make people think:

  • This 1.0 demo kludgy simulation is better than a poke in the eye with a sharp stick—it actually is useful in some cases.

  • We can now see what the version 3.0 of this product will be.

  • Building 1.0 impresses us that the team is competent, pointed in the right direction, and has the resources to build the 3.0 version that we would want.

  • Therefore we are willing to become ground-floor early-stage supporting investing customers for the ultimate product.

Get 33% off a group subscription

I am reminded of Bill Janeway’s explanations of how “Enterprise Software” used to work back in THE DAY, with contrasts to how modern VC works. What you would initially ship was something for which you would then offer a:

Bill Janeway: The Rise & Fall of “Enterprise Software”: ‘Perpetual, non-exclusive license paid upfront in cash by the customer for code that was inevitably incomplete, a work in progress, which came with 20 percent of that license fee due annually so that the vendor would keep the stuff running as the environment in which the code was running continued to evolve…. The financial consequences of this model were absolutely terrific. It meant that the rich customer was funding the poor startup with no issuance of equity, no dilution of ownership, and of the upside rewards available to the investors and the entrepreneurial founders of the business—the cheapest capital imaginable… <https://businessofsoftware.org/2019/08/rise-fall-enterprise-software-bill-janeway-warburg-pincus-bos-europe-2018/>

Share

Perhaps this is what Apple is trying to do with AVP—but on the “creator-retail” rather than the enterprise level. Perhaps this is the track that the people behind the HAIP hoped for it to follow.

There have been a number of ways of trying to get first cogs and gears, then electric relays and switches, then vacuum tubes, and now crystal sand to get up and dance in sync with human thought: the Difference Engine, punch cards, MEMEX, FORTRAN, UNIX and its command line, Xanadu time-sharing, Windows-Icons-Menus-Pointer, DynaBook, the PC, Apple Newton, Workstations, General Magic, Blackberry, Palm, the iPhone, Oculus, Google Glass, Alexa, Modern Advanced Machine-Learning Models (MAMLM), and much more. And next… what?

What determines which systems get their chance, rather than being cut off on or before launch? And what determines which systems that can launched actually find a computing-power and human-accessibility fit?

One thing that does seem to be becoming increasingly clear is that venture capital AG (after Google) does not serve humanity’s interests in technological exploration and development very well:

Benjamin Sandofsky: Oh the Humanity: ‘Humane raised $230 Million and spent six years to ship an “Ai Pin”… [that] shipped with fraction of the features promised… that… fail most of the time…. Too heavy… atrocious battery life… projector… useless in daylight… input system… awkward and slow…. [But] this isn’t the story of a young man given too much money and power…. Humane’s founders worked on the first iPhone…. In 2018, there was more money flowing into Silicon Valley than there were quality startups to invest in. That’s OK for VCs because they actually make money whether or not they pick a winner. They charge “maintenance fees” in the range of 1.5% to 2.5%, so when an endowment writes a VC a check for nine figures, the firm makes millions a year. This meant that Humane didn’t have to be a knockout, they just had to be the last single person at the bar at closing time…. When interest rates went up in 2022, startups suddenly found it harder to raise money. Investors were suddenly interested in—GASP—profitability…. Humane needed to ship something to look like a later stage company, but as we now know, their hardware sucked. They decided to latch on to AI…. The problems with Humane began at their seed stage. They should have tested the idea faster, and rejected it right away. I get that hardware is different than software… <https://www.sandofsky.com/humane/>

Share Brad DeLong’s Grasping Reality

The requirements of the venture capitalist ecosystem today made it all but certain that the HAIP team’s work would not be well-focused. Of course, neither Microsoft Research’s work back in THE DAY nor Google Research’s work today is well-focused either…

AVP is going to get its chance to turn from a technology demonstration and simulation into a product. The HAIP is not. Is that how it should be—that the first is actually much closer to being a product than the other? (Bear in mind that developing the HAIP has so far cost $150 million, while developing the AVP has spent… only the Storm God of the Semites knows how much to date…) My view is that the eyes—glasses—are an interesting place, the wrist—watches—are an interesting place, the hand—smartphones—are an interesting place; but that the chest-attached broach is not so interesting.

But what do I know? Not much.

Leave a comment

References:

Subscribe now