This is a really tough one for me to make. I am a software engineer at Microsoft, working on Copilot. I am someone who’s always been extremely passionate about bringing bleeding edge tech to the forefront in apps I make. I want the power of AI to be brought to the masses.

The problem is, every single one of you in Silicon Valley is thinking the exact same thing.

That’s how trends work. It’s not as shallow as I originally thought. “Make AI slop?” I thought, “Why would I contribute to that ecosystem? I’m making AI for the benefit of humanity. The thing I’m making actually matters, and is better than those other things”. This has been my philosophy. I didn’t think I was riding the hype wave, I thought I was breaking free by applying my unique, fresh take to the overpopulated and bloated AI ecosystem.

Except, everyone is thinking like that. Even at Microsoft. Other people are doing this, but we’re X and we’re going to do Y to differentiate ourselves. Except what you end up with is the exact paradigm you were trying to avoid in the first place – slop. We’re seeing this a lot in the startup world. Companies make “the next big thing” and it’s a ChatGPT wrapper. Again, even corporate is very guilty of this…

Which has led me to think – if everyone is trapped in this endless cycle of chasing – when do we actually get the thing we’re talking about? AGI, Grok will bring us to Mars, everyone can fire their secretary for a robot that works for free – fantasies aside, what is actually being made right now? What’s the reality here, is it innovation, or just empty promises?

The reality is actually empty promises. Practically every AI company is actively losing money. OpenAI is not profitable. They are running a SaaS that touts itself as a genie oracle, so of course the world is going to take advantage of their systems to get maximum utility. NVIDIA and the people providing the hardware for these AIs to run on are the real winners here. They’re just selling what they’ve always sold to people running off to chase this “AI” thing.

The complexity of software

All of this ties back to what I think is the bottleneck here, the inherent complexity of software, and engineering for that matter. The simple principle that, it doesn’t matter how big your systems are, complex problems require exponentially complex solutions. However simple you think your favorite website to code is – actually is eons more complex than you could even imagine if you’re not an engineer.

Take real buildings, for example. You can’t just go out there with the materials and no plan and, bam building. You need a massive amount of prep – not only with the materials you’re going to use, but to account for weather, decay over time, all of these things you just expect as a person inside the building to not have to worry about, but as the engineer you absolutely need to worry about – because if you don’t, the people in the building will! And that can’t be good!

The same applies to software – it looks really easy when you boot up TikTok and get videos slammed in your face. But the reality is there’s so much more going on behind the scenes. To even get the recommendations loaded that fast tailored to you is, pretty impressive actually. We take a lot of this stuff for granted because, that’s what people do.

With AI, the thing that became super apparent to me that it actually isn’t meant to be a product is, we in our entire existence as humanity have never had to worry about a “second” form of life or sentience. AI is the very first paradigm to challenge our existence, or at least morph a part of our existences out of the picture. Computers definitely already did this for more laborious tasks, but even critical thinking tasks like coding now can be automated with AI. This thing that we barely even understand has begun to take over our functions as humans.

With this said – how the hell are we actually supposed to use this? Like, what are we to actually do with this information? Literally the only interface we’ve conceived of is the chat interface which when you think about it is incredibly stupid and lazy. “We don’t know how to best utilize this incredible technology so you can just chat with it like you can chat with your friend haha” – or, something like that. Look it’s not to downplay the impressiveness of AI, it’s an insanely incredible technology with lots of insanely smart people making it work. It’s just that… we don’t know how to use it. It would be like if we suddenly mined up Vibranium and were like, cool, really awesome, but what can we actually DO with this? Sure a “Vibranium powered world” sounds really cool, but how is it actually made, and is this something that will ever be done?

What is AI then?

If AI isn’t the product, what is it then? I don’t know, Javascript isn’t a product, what is Javascript? Certainly Javascript only powers every website in existence, so it’s definitely useful. But, useful things don’t have to be the end product.

What if AI was just a tool that let us build the useful things? What if, we’re wrong about what AI is? It’s not the end goal – websites don’t all need to turn into these lazy ChatGPT wrappers. We don’t even need to come up with some new radical new interface that lets us maximize the output we get from AI. What if we just, use it to make cool things? Things that are actually useful, that provide real tangible value?

It’s not to say that AI will never provide real value, in fact it already does to an extent. But, it certainly does not provide the amount of value that it advertises. ChatGPT will maybe help you on your math homework every now and then, or help you understand something about your tax form that Google can’t quite explain. ChatGPT will not give you the will to live, teach you how to love, actually do stuff for you, the list goes on to infinity.

All of these “out there” pitches for redefining the UI of AI, trying to force people into this silo they don’t want to be in to me feels… wrong. We should be optimizing the workflows for people, not forcing people to use workflows that WE say are optimized. Here’s a good example – did anyone watch the Coinbase ad yesterday for the superbowl? Because, it was really good for about 95% of the time! Then it ended with Coinbase and everyone about lost their minds with the colossal wave of disgust that hit them when they realized it was a crypto ad (not only at my place but, the reactions online are telling). Crypto will never be mainstream for the sole reason that nobody wants it to be mainstream. It’s not that it shouldn’t be, it’s that it realistically will never get over the image that millions of people around the world have already formed about it. It had its chance, it failed, maybe wait another 100 years until everyone who remembers crypto from their lifetime dies?

Nobody wants anything shoved down their throats. If online sentiment from AI is at all telling, it’s that people are pretty clear on their boundaries between what AI can and can’t do. And everything being a ChatGPT wrapper is something I’m pretty sure, on average, people don’t like.

Again it’s not to say that AI will never be a good product, I think ChatGPT, Anthropic, and some other guys have decent enough reputations such that they could release the ChatGPT phone tomorrow and everyone would buy it. It’s just that, time might be running out for this to happen. Everyone has already been complaining about all of the AI Superbowl ads, I cannot imagine our country taking another year of AI Superbowl ads very well.

What now?

The question I’m sure millions of us are hit in the face with. If AI isn’t the product, why is everyone freaking out to make sure it is, and when will it stop? I see either two breakpoints to this reality playing out:

1. AI does become the product. A profitable one, though. Unlikely, but possible.
2. AI does not become the product and the economy explodes

Either way, AI as a productivity tool – that’s not going anywhere. At least, I can’t tell you a single software engineer who has used Claude Code who will willingly give up Claude Code for… anything, really. It’s just too transformative in the builder realm. So in that, it’s changed life for the better.

For everything else, only time will tell. But don’t put all of your eggs in the AI basket just yet – the winners of the AI race might not be the ones who build the next AI, the winners might be the ones who use AI to cross the finish line.