It didn't take a rocket scientist to roll their eyes at the stupid "Humane AI Pin" well in advance of its release. It was like they were reading from an AI generated script about a senseless product that was never going to work, following in the wake of so many other scammy gadgets trying to claw as much funds from your wallet as possible. Yet despite massive high profile blunders from Google to Humane, companies continue to pile resources onto the AI bandwagon and continue to fail miserably.

Will Apple's AI Flagship Flop?

Let's be real: the Vision Pro was a dumb idea. Just like the "AI Pin", it was a struggle to square the novelty of the mixed reality world with the absurd price tag. Anyone that's purchased a VR since the advent understands very well how this tech works. You strap on the headset and are overwhelmed by how damn neat it is. Then you take it off and use your computer like normal because as neat as it is, immersive experiences are not inherently better than so-called "flat" media projected on a conventional screen.

Some "people" (read 'CEOs') don't get this dynamic at all. They strap on a VR headset and believe they're stepping into the future because they don't have the attention span required to actually use these products the way consumers do and realize that it's easy to become fatigued with so-called "spatial computing". I don't want to physically pinch windows or Minority-report interfaces around in some exhausting hand-dance designed to look good for a freakin' movie, not to be a practical interface.

In other words, the Vision Pro wasn't a great idea at its core unless they found a definitive way to combat this spatial computing fatigue in a way that was truly, truly as good or better than conventional experiences. A massive, $3000+ weight strapped to your head for hours on end proved a bad idea.

There's an old saying, "All Politics is Local". So it is with corporate behemoths, too. Tim Cook needs a big win because a mix of pride, ego, and external pressures. If he keeps churning out duds, he's not going to keep his job.

Let's Talk Efficiency

I've beaten this horse to death: AI isn't an efficient tool. It can understand human text better than any other artificial system ever created, but it requires buckets of energy to do what our brains can manage with a few piles of corn powering our bodies. It's not efficient at all, even if it is fast.

For a phone that is already rather infamous for overheating, Apple has announced that their next generation of software will "empower" your device with new AI tools (at least for those with the latest model, of course), some of which run locally on your device. They pitch this as a selling point, but for anyone that's worked in (even the shallow end) of this space, you'll understand they did this because they didn't want to shovel cash at OpenAI's usage based API.

They will still end up doing this, which isn't a problem for Cook. Apple is infamous for having mountains of cash reserves and Cook's legacy is more important than anything else if he wants to keep his job -- Apple just doesn't work if the perception is that their CEO is out of touch. Image is everything for the company that still thinks they "think different".

Why will they have to rely on APIs, ultimately? Because there's no physical way to squeeze the awesome amount of resources LLMs consume onto a phone without heat as a byproduct.

For a phone that's already had struggles in managing heat, how will this work...? How do you prevent the local LLM from chewing through battery like a starving man at a Vegas buffet? LLMs are energy vampires and I encourage you to spin up Open Llama and see this fact for yourself. Even assuming Apple aggressively quantizes the model and relies on the cloud to run more complex queries, this has a lot of potential to go very poorly.

AI Products Suck because CEOs Don't Understand It

The greatest potential in AI is that it is the first conduit we have to communicate with digital constructs "directly". It's power is in natural language processing, not in being a chat bot. That's not a bad application, don't get me wrong, but the power of AI isn't that it replaces Google in finding you the right answer. In reality, finding the right answer is arguably its weakest attribute. The power isn't in that; traditional algorithms are probably superior with this, still, as anyone that's used GPT will know.

The power is that for the first time ever, we can ask a computer something and the computer can reply without the traditional middle-layer of a computer programmer. That's all programming really is: translating human thought and logic into something a machine can understand. The ability to remove that translation layer and shove your query into a neural net which the software can then "understand" is a massive superpower that proves that AI isn't mere hype...but we live in a world of capitalism where an idea's worth is born through products, not usefulness.

The big corporate players seem to dance on the bleeding edge of tech, vying to show their restless investors that they will never go the way of MySpace or Yahoo. Yet that's exactly the danger when they push for tech absent understand how it really works, instead focusing on things like personal legacy and this "race" to prove relevance.

The problem with Apple's AI integration is that it is destined to suck. Even if it works exactly as well as GPT4-O embedded all over the OS, are you sure you want to drain that much battery for some pointless little operation that will probably not work...? Do you really want to expend all that extra heat and battery just to get back a hallucination? The "generative" facet of generative AI is actually the least interesting facet to me, it's the idea that the computer can "understand" my language directly. That's where the potential really is.

If this goes poorly, it will even incentivize people not to upgrade their phones or (*gasp*) switch to another platform entirely. I'm not going to say there's zero potential with Apple's vision (pro), but I've yet to use GPT 4 and think "hey, I wish I had this embedded on my phone".

With reports showing that average users are using GPT less and less, I wonder what value proposition Apple is actually offering with these tools. I'm sure an endless stream of adverts will tell me, but I somehow doubt I will see it...and I'm skeptical that all but the die-hardiest Apple fans will see it, either.

Of Course AI will Crash...or will it?

Investors are not the smartest people in the room...they are just as diverse a group as any other. The only defining attribute of an investor is that they have capital, and anyone with a brain knows that simply having capital doesn't make you a genius.

It doesn't take a rocket scientist too see that the AI space is overinflated with startup ideas that truly sound like they are AI applications churned out by AI. Many investors don't even know what AI is, often confusing GPT-like neural nets with classic machine learning and failing to grasp how GPT is wildly different from things like a recommendation engine.

Of course, the strange reality of this world is that a product doesn't actually have to be useful to be successful. Consider Bitcoin...it's absolutely not a currency, but a financial commodity valuable only because investors say it is valuable. As actual currency, it's mostly useless due to immense volatility and lack of accessibility, but it has value because investors say it has value. The grand ideals about Bitcoin as a decentralized currency never materialized and never will (despite what some people still insist), yet the widget still has value because we live in a world of investors, not a world where a product has to be useful to be a commodity.

AI isn't like Bitcoin, though. It isn't a commodity in the same way and ultimately the wave of AI investment has to crash against something, and it won't be massive mounds of profit because most of these startups aren't working on anything useful or practical that is actually a good fit for the tech.

Conclusion

There will be a lot of failures in the AI space, and I'm not sad about it if it means rehiring some of the great engineers that companies have decided to let go despite record profits and prosperity. For CEOs and Stakeholders drowning in AI FOMO, really think about the risk to your brand. Google seems to think that it doesn't matter if their AI product is shit, so long as they are first to roll it out and not be displaced by some upstart AI search engine.

That's a fear-based reaction, not a conscious, deliberate step to improve their product. Tech companies that were once upstarts like Google now fear their own irrelevance, ironically degrading their products because they are no longer focused on building value for their consumers, but for themselves. Is this AI push really about what will work best for Apple users...or Tim Cook's stinging failure with Vision Pro and the cultural drive to be "cutting edge" no matter the cost...?