Winter Rant

The past six to eight months have been a study in contrasts, in tech. Roughly eight months ago (may have been earlier), OpenAI put out a simple looking demo, backed by incredible engineering and technology, but little to no product finesse. ChatGPT was a roaring success. It heralded a new age in AI, a new age in software, and a new age in technology. Businesses contorted themselves overnight to accommodate for an AI chat bot (if not entire armies of such bots) in their daily operations.

Contrast that with Apple’s week-old announcement of the Vision Pro. It is an expensive technological marvel in a space (virtual/augmented/mixed reality computing) that was all but written off, seemingly. Apple had to call on every product and engineering know-how that it gained over decades of shipping consumer devices to land on the most polished version of a mixed reality headset with a compelling software and user interaction story. Apple touts a lot of arcane features and improvements when releasing iterative product versions in its other offerings (watch, mac, iphone, ipad, airpods, homepods). I would normally not give those features a second thought. Consider the amount of engineering that Apple went through in the headband of the Air Pod Max, to deal with the weight issues of those heavy headphones. I could see echoes of such prior craftsmanship in the Vison Pro. Apple essentially funneled every little nugget of product expertise into the Vision Pro to great effect – the end result looks great.

But it took all that product and engineering finesse by Apple, to land on what seems like an MVP (minimal viable product) from a business standpoint. In contrast, OpenAI’s ChatGPT was hardly groundbreaking in its ‘product’ thinking, but it shook up the commercial and business landscape. As an immediate data point, Apple’s stock was down after its Monday announcement of Vision Pro – that tells us what the market(s) thought of Apple’s bet on mixed reality, in a moment otherwise dominated by AI.

Interestingly, both ChatGPT and Vision Pro are groundbreaking technology successes – we certainly live in a new age of computing.

All of this though, has me thinking about a comparison of a different kind: my own emotional reaction when I first saw and read about these announcements. Now there is a bias for sure: I have tried and used ChatGPT, but not Vision Pro (given their lopsided availabilities). But I have been thinking back to my emotional reaction to ChatGPT even before I signed up for an OpenAI account.

The idea of a machine learning model that is good at transfer learning – where the ML model trained on one data set and/or task is applied on a new task – instilled a sense of dread in me. There I said it. I am a software engineer, and AI has been the holy grail for software engineers. I like nothing more then automation. If there is a task I can delegate to a bot, I would love nothing more. But my initial reaction to ChatGPT – a model that is particularly good at transfer learning – was that of fear.

Again, in contrast, my reaction to the idea of spatial computing – where digital representations of the real world can both exist and be manipulated with code – has always been that of amazement and wonder. And this has been my reaction ever since I saw the Google Glass announcement more than a decade ago. And those feelings of limitless possibilities reemerged with Apple’s announcement earlier this week.

Why would AI scare me, but AR/VR inspire me? There are a few answers ricocheting in my mind.

At its most benevolent and virtuous self, artificial intelligence’s end goal is to mimic my abilities as a human, with the possibility of replacing me at tasks that I can otherwise perform (manually) as a human. At its most benevolent and virtuous self, mixed reality’s end goal is to overlay my senses and environment with useful information that extends beyond the 2-dimensional nature of a computer screen. Let’s just say that I do not like being replaced.

Next, I need to consider the opposite end of the benevolent/virtuous – malice and vicious. For that I outsource my imagination to Hollywood – the ultimate ‘dream machine.’ Movies have already painted hellscapes for both these technologies. I give you Terminator and Matrix.

In Terminator, an AI program takes over the planet and nukes it. In Matrix, humanity is subjected to an endless loop of virtual reality, so that it could be an energy supply for its robot overlords. Even in those worse-case scenarios, I rather be plugged into an endless loop of virtual reality than be nuked. Although when I think about it, the robots did nuke the planet in the Matrix as well – I am sensing a theme here.

As a programmer, AI and spatial computing offer new paradigms to write code – so I am equally excited about both spaces. But I think there is only so much damage that mixed reality can cause. When a mixed reality headset fails, I just reboot it, or take it off entirely – I am still in control, as a human. When an AI goes off its rails, I cannot control it, because I cannot explain why it is “hallucinating.”

I suppose it comes down to who is in charge, when drawing out humanity’s relationship with products coming from both these spaces. Vision Pro or anything like it, is just another computer in a different form factor. Humans are still at the center of that form factor. With AI however, the human can be ignored.

– vijay, enjoying some mild weather for a change ⛅️