We’re Living Through Sci-fi

I try to live in the future; it’s just better. Many ponder the period in history when it was best to be alive. The answer is always now. Compared to many on our beloved planet, I am fortunate enough to live as close to the future as practically anyone. So, forgive me if I go a bit futuristic today.

But before all that, let me start in the here and now. If you are a developer, you know GitHub Copilot. If you’re an artist, you know DALL-E 2. If you are a blogger, you’ve used ChatGPT. Even if you are none of those things, unless you’ve been off grid for the last year, you’ve still probably heard of recent innovations in artificial intelligence. As everything needs a snazzy buzzword, Copilot, DALL-E 2 and ChatGPT are called generative AI. VC money has poured into generative AI startups, talent has flocked to build the latest large language models, and people proclaimed that the future has arrived. We should be cautious about proclamations of the future. Remember the Sinclair C5, 3D at home printing and Web3?

But perhaps, this time, it really has.

We’re now able to, quite remarkably, write a prompt, such as “a 1000m megatall skyscraper, with an architectural style similar to the Shanghai World Financial Center, with orange LED lights, showing the full skyscraper”, and have images generated. I’ve just tried this prompt.

My god, I love DALL-E. As a child, I occasionally enjoyed drawing skyscrapers, their size being the most important thing. I was a frequent visitor to skyscraperpage.com. Back then, I had to draw - on paper! Now I use DALL-E 2.

I live in continuous amazement that this is possible. We’re living through sci-fi. My life is now a version of the tech seen in 90s action movies. Back then, biometric scanning only worked for James Bond, now it works for all of us.

Technical blueprints for production lines, planes and ships; database schemas; code to build websites; supply chain designs; ML pipelines. All will be generated from large language (for text) or diffusion models (for image generation). The world will be transformed. The productivity growth could be astounding, and GDP will grow.

We’re in the midst of lowering, or entirely removing, barriers to entry to many fields, many skillsets and many professions. It’s a genuine moment in time, an epoch in history. All we have to be able to do is write a prompt.

At the risk of sounding like a failed prophecy from Tomorrow’s World in the 1970s (why don’t we have flying cars), generative AI will change it all. But we haven’t seen anything yet, because there are three extensions which I’m eagerly awaiting.

Whilst I’ve created artistic drawings of skyscrapers, we will be able to create detailed technical architectural designs of skyscrapers. The time to create a blueprint, which adheres to the laws of physics, will reduce to effectively nothing. Do I want to build a skyscraper - call it Mark Tower - yes, I do! Do I want to learn materials science, architecture and the principles of physics to do so? No. Soon, I’ll be able to design all that while having a cuppa tea at my desk.

Sometimes it feels like our imagination is richer than anything we can type or speak, which brings me to the second extension. I swear my blog posts always sound better in my head (insert your sarcastic comment). With current generative AI, as remarkable as it is, we have to think, then prompt: “a function that replaces null values with 0” or “a birthday greeting in the style of a haiku.” Imagine the power and the productivity if we don’t have to prompt. Imagine if we only had to think! The brain-computer interface, the next level of human and machine symbiosis, is an approaching frontier. It’s all a bit Matrix-y. Elon’s Neuralink is researching this heavily. I’m excited to see how far it can go. I can simply think of my skyscraper, and then DALL-E 14, or whatever version we’re on by then, will create the blueprints.

The last extension is the most out there. Can we have generative AI for hardware? Self-replicating robots. Or what I’m calling generative hardware. Von Neumann explored this area of research for much of his later life. If it’s good enough for him, it’s good enough for me. When machines and physical objects can replicate themselves and create more machines, we will see a revolution like no other. The speed of building a skyscraper would rapidly reduce. The time to build a dam, a road, or a railway decreases to weeks. When we return to the Moon and finally get to Mars, self-replicating machines will enable us to develop habitable spaces much quicker. We’ll no longer have to wait years to achieve economic gains from infrastructure projects (HS2 says hi).

If this sounds a little too far fetched and you can’t envisage construction sites which employ only robots, imagine a person from 1973 seeing the world today; technology has changed our world almost beyond recognition.

The power to harness the full extent of human creativity is exciting. AI lets us program and build using natural language. Sam Altman wrote, “AI is going to be the greatest force for economic empowerment and a lot of people getting rich we have ever seen.” I agree. Generative AI allows anyone in the world, anywhere in the world, to start building websites, to begin writing and designing, and to build businesses.

Thanks for reading, I’ll see you next time!