AI vibes

‘Vibe’ is the hot topic in AI land. You would think that it came from the Beach Boys' Good Vibrations, where everything that has AI attached to it is just a bundle of happiness. Good guess, but no. It comes from the new hip term ‘vibe coding’, which was coined by Andrej Karpathy, a co-founder of OpenAI.

The idea is to describe in human language what you are trying to build and with a press of a button - and some magic behind the scenes - the AI will return the required programming code. As a developer - the assumptions go - you get relieved from the mundane task of writing code and instead of that you can  focus on high value work.

News like that combined with similar announcements (“Amazon Q Developer: The most capable generative AI–powered assistant for software development”) give the impression that the job opportunity for developers had its day in the sunshine and that it’s a profession past its best-before date.

The irony starts with the fact that a number of companies getting into this automated coding is growing and they can't hire enough developers to build these tools. One would think that getting to work on a tool which will make you obsolete would be the last thing you want to work on, ever. I am sure the developers  think it doesn't concern them since they can code better than anyone else, even better than AI.

The next argument - and very, very shortsighted - is that we will no longer need junior and mid-tear developers. Just a few senior, old dogs, who, with the help of this new tech, will produce so much code that computers will have a hard time using it all.

Of course, nobody is asking where we find the next batch of old dogs, who had the chance to work long hours to master the craft. Even worse, universities are promoting computer programming courses where the value proposition is that you don't have to worry about the details, and you just build apps to solve the world hunger problem.

Not to be outdone with perfectly logical arguments, we are told that the non-technical people will be able to start building their own apps in droves. That very well might be.

As a hobby, or for your own amusement, you could ask AI to build a simple app to tell you when you are hungry and what and where you should eat to remove the burden of decision making from you. In case the app crashes, you are still left with the option to go to your fridge and see what's there. You can also imagine enterprise settings where unleashing such creativity might not go as well as planned.

Here is a standard test - look at people next to your cubicle or on the adjacent squares of a Zoom call and ask yourself if these people  can even order at the drive-through window. And these people will be telling AI what to build?

The part which is missing in these narratives is that writing the code is one of many activities required to deploy a fully functional product to the end users which meets their needs. A code, which is secure and maintainable.

Let's assume for a moment that this will work. It is Monday morning and your company is left with a few senior developers, who start producing millions of lines of code just by thinking about it. Actually, do you really need them when you have your product managers who suddenly don't have to talk to the developers and are free to just describe what exactly they need the software to do?

Actually, who needs project managers, when the people in the marketing department can unleash their creativity and, in technicolor, describe the dream product to AI? Before lunch is over, the product is launched and the company CFO is mesmerized by the amount of money flowing into the company bank account.

We also know that bean counting is one of the professions which is totally replaceable by AI.

And that takes us to the interview where Sam Altman, CEO of OpenAI, mused that 'In my little group chat with my tech CEO friends, there’s this betting pool for the first year that there is a one-person billion-dollar company.'

Enchanting vision. But before we get there, Mr. Altman needs another $40 billion to keep OpenAI afloat. He also needs OpenAI to start generating $100 billion for Microsoft before OpenAI is released from the shackles of the contract with Microsoft and can flourish on its own. By his own predictions, OpenAI won't be cash-flow positive before 2029. 

To further demonstrate how incoherent its strategy is, OpenAI is in talks to buy Windsurf, an AI coding company, and is contemplating building its own social network. Purchasing Windsurf (Codeium) would provide another possible stream of revenue and justify its sky high valuation.

Here is my opinion.

All the talk about ‘vibe programming’ and replacing people with AI is a distraction. Talking about a $1 billion company created by one person is similar to the idea of selling a book titled 'How to win the lottery'. If you are a CEO of a company with the most advanced technology, the best scientists, which has access to the whole Internet, you should be able to start building thousands of $1 billion companies.

So far, this company can't even get the pricing for its products right and it is losing money just by introducing new, better products. I am sure it will make its money on volume.

OpenAI is trying hard to create products which would bring sustainable, profitable revenue to the organization. Its (incoherent) strategy is to create an 'AI Portal' where all your digital needs would be met and you would never leave for any other platform. The time, money and resources are not on the side of Mr. Altman. The vibes are not good.

In order to have a defendable strategy, you need a cohesive story. OpenAI doesn't have it. And that's the recurrent pattern.