Interview: GoodData.ai CTO says the enterprise AI bubble is real, but so is AI’s transformational power

The question of whether an AI bubble exists has now turned to if, when or how will it may burst. Signs include early-stage AI startups with valuations previously only seen in established tech giants. A handful of companies control the narrative around the productivity benefits of enterprise AI. And organisations that rushed to implement AI with massive investments over the last two years are beginning to scrutinise ROI. What will be the aftermath of this inflection point?

Peter Fedoročko, CTO of GoodData believes two things can be true: He acknowledges that there is an AI bubble, but insists the technology still has real value. After the dot-com crash, he says the internet didn’t disappear, it became infrastructure. And he expects the same thing with AI.

Peter Fedoročko: “The dot-com crash took a decade to recover financially, but the internet reshaped everything during that time. It didn’t wipe out jobs, it transformed them. AI follows the same pattern. Once the hype burns off, the real builders get back to work.”

Lara Williams: Isn’t the analogy between the dot-com crash and the AI bubble a false equivalency? The internet was infrastructure. AI is outsourcing cognitive work. Why do you think the outcome will be the same?

PF: “Every analogy simplifies, of course, but that’s what makes them useful. The dot-com parallel isn’t about them being identical technology, because they’re not. Far from it. The internet was infrastructure for communication, whereas AI is infrastructure for automation. The analogy I used is about identical human behaviour. We overbuild, overhype and overcapitalise, and then the market eventually corrects. The functionality and utility of any groundbreaking technology will always survive the market correction if the technology is real.

“The internet had utility. So does AI. The outcome will be the same because AI will transform people’s lives just as the internet did. The way we operate now, our kids who are growing up in this era of AI, well they will look back at it and be in shock. Just as we did to our parents who did everything pre-internet. It’s not a case of false equivalency, the analogy is pattern recognition.”

LW: But the internet created the foundation for a new kind of knowledge economy. What kind of economy is AI creating if knowledge work is outsourced to machines?

PF: “The internet created a new knowledge economy and also externalised memory. Search engines have long since replaced the need to remember phone numbers, addresses, and random facts off the top of our heads. And yet no one calls that the end of thinking. Everyone looks at that benefit as leverage. AI is a further step along the same continuum.

“The question was never whether machines would take cognitive load, it’s whether people will use that freed capacity to move up the value chain. I don’t think we will live in an economy where AI is doing all of the work. It will do the boring parts of our jobs, the parts that our brains are too valuable to spend time doing, thus freeing up people’s time to do the parts of the job that they love – the complex parts, the parts they have been trained to do. History suggests that we will adapt, and people will move up the value chain. Look at the Industrial revolution, that completely changed the way we worked, and the sorts of jobs that were available. It might not be quick and clean, but we will adapt.”

LW: If you are right about AI being the infrastructure of a new economy do you have a line of sight on what part of the cycle we are at right now?

“We are somewhere between peak inflated expectations and the trough of disillusionment, to borrow Gartner’s language. The infrastructure is being laid right now, during the chaos of this AI phobia society seems to be developing.

“To draw on another comparison from the early 2000’s, we can compare all of the groundwork for AI infrastructure to the laying of the fibre-optic cables. They were being laid while pets.com was going bankrupt. The difference is that most people couldn’t see the actual infrastructure that would make the internet transformative. They only saw the stock prices collapsing and declared the whole thing a fraud. But the cable kept getting laid, the people laying the infrastructure didn’t care about the sentiment, and that’s exactly what’s happening now.

“The compute is being built, the models are improving, and the use cases are compounding. AI has already changed how code and design is produced. To borrow the silicon valley terminology, the future is here – it’s just not evenly distributed. Yes, there will be companies that fall, the companies with no clear revenue model and a chatbot bolted on as a product will obviously not survive. But the underlying capability of the technology? Well unfortunately for those with AI phobia, that’s not going anywhere, you can’t uninvent intelligence”