I write this at a point where the company Ankita and I started is no longer relevant. It has been run over by the latest image generation model of ChatGPT.
Our company ImageTranslate basically let anyone create a different language version of an input image using a bunch of clever tricks of image processing, natural language processing and UX. In 2019 this was quite an challenging thing to do. As of March 26 2025, this is trivially solved by a single prompt in ChatGPT. So yes, while we scamper to figure how to tackle this onslaught of AI models and pivot our business, the epitaph of our company as it exists today has been written.
As a kid who’s poked around computers since the Windows 85, Pentium 2 era writing my No-CD cracks in assembly to the current stint as a product mananger I don’t want to let this latest tech-train run over me. It’s already caught up too much speed as I’ve lazily watched it from the platform. But it’s time to make a run and catch it or risk obsolecence.
I don’t know if you know but the number of bigwigs who have openly announced that ‘learning to code is no longer important’ is unbelievable. The much revered world of tech and IT is going to undergo a metamorphosis that will be a sight to watch.
Companies will need 1/4th of the current workforce to get the same work done. Fewer engineers, product, design, ops, sales, everything. The bloodbath hasn’t yet begun but it will. Humans generally respond slowly. My intention here is not portend an ominous future but to atleast ponder it.
Let me be honest. While I watched the rise of the LLMs since the first version of ChatGPT came out, I didn’t move a finger to build something using it. I’ve only and only devoured news and engaged in lunch-time conversations masquerading as a knowledgeable guy. But let me admit it. I know nothing about how to create value on top of the AI world we live in.
To ponder about the future, I must however first look backwards to atleast have a system of thinking. There have been a set of revolutions in human history that had an outsized impact on the future.
-
Agricultural revolution that harnessed the abundance of natural resources: land, water, livestock, sunshine
-
Industrial revoution that harnessed the abunance of energy: thermal, electric, chemical, nuclear, solar
-
Information revolution that harnessed the abundance of data and computation: internet, storage, GPUs
The current stage of AI seems to be the culmination of information revolution. Ofcourse its not exactly the year 2025 but yes, +- 10 years will be considered the inflection point.
We already live in a world where every single piece of information recorded in the past is available in an instant. And every possible of variation for the future can be created in an instant. Text, images, speech, video, 3D. Or in colloquial lingo: A presentation, spreadsheet, email, code, cartoon, lesson plan, poem, advertisement, love letter, marketing copy, design, novel, hate speech, itinerary, web-page, video game, balance sheet, exam, thesis, test case… whatever that can be expressed in language into a computer or a notepad. Whatever! That too at a degree that surpasses the median human work by magnitudes.
What it lacked for a while was agency. The ability to actually get things done. It would write code, but wouldn’t run it. Would make a design but not implement it. Would draft an email but not send it. Would make cartoon, but not publish it. Would make an itinerary, but not book it.
This is changing fast however. Ther are interfaces being built that lets AI speak to existing tools to simply get things done. It can book your flights on Skyscanner, update your balance sheet on Quickbooks, commit code to your Github repo, run your ad on Instagram. Pretty wild.
This is very empowering but also scary if you watch it in action. If it does something when you ask it do it, it’s empowering. Scary if it started to act on its own. A self-driving car is supposed to get you from A to B so you partly know its intentions. What are the intentions of a self-driven AI? Or are there any at all?
If ChatGPT could use your computer without any instructions from you, what do you think it would do? Empty your bank account? 😅
TBH I’ve gone down a rabbit hole that doesn’t have clear answers yet. AI/models built by big corporations act depend on what their masters programmed them to do. Right now the masters seem to be the end-users. But this argument isn’t eactly a detour. We need to make fundamental assumption that AIs will be servile to proceed with our quest for the future.
Given these omnipotent-but-servile AIs, the main question then is about what we do with them.
-
Use them This is such a powerless thing to say but it needs to be said. An extremely big section of the world’s population still does not seem to use AI. Even the poorest human being on earth, if they atleast have access to a phone and internet, suddenly has access to a personal servant. That too a very wise and intelligent one. Why would you not use it? Abstaining is akin to using candles after electricity has been invented.
-
Get comfortable It is intimidating to have an assistant that is wiser than you. Let me admit it whether you want to admit or not. Turns out this is fundamentally similar to having a sibling, friend or a colleague who is smarter than you. The natural thing you initially feel is intimidation but the wise people eventually realize that having these people around is an amazing chance to learn and soak up knowledge. By the way, if you use it enough you will notice that it does make mistakes which at times is very comforting.
-
Don’t be too comfortable Your job is under threat and so is mine. The AI’s mistakes are decreasing and its proficiency and agency increasing. Employers have already started to notice. How can they not? Forget jobs, the company itself is under threat in certain cases. Anticipate difficult times and start preparing for them.