Last week, The Massachusetts Institute of Technology has published a report This would make all the CEO sit straight: 95 % of the immunic intelligence pilots achieve a zero return on investment. Allow me to put it in the right quorum, the startups of artificial intelligence took about $ 250 billion in adventure dollars over the past 12 months, and only 5 % of these projects have created measurable value.
This is the definition of gold rush. Investors throw drivers to anyone who is in digital gold. But most of these prospectors do not know what they are digging, not to mention how to build something that lasts.
I manage a robot company This builds programs that operate in Amnesty International for some of the most difficult industries on the ground-energy, defense, manufacturing and mining. We have been equipped for years and arrived Unicorn mode Just a few months ago, with a value of $ 1.25 billion. Everywhere I was going to, whether it was Davosand Washington, DCOr The Middle EastThe story is the same: CEOs have been sold in reality. The noise is real. The basics are wrong.
Why do most artificial intelligence pilots fail
Here is the difficult truth: Amnesty International does not fail due to weak algorithms. It fails due to poor data.
All Fortune 500 Boardroom has seen the stadium. Elegant chip floors. Dazzle experimental offers. Comprehensive promises about the transformation. But scratching the surface and you will find a deadly defect, the companies do not have the data that these algorithms need to work.
Data quality is poor. Lost data. No earthly truth. This is the real problem, not mathematics. The algorithm is as good as the fuel you put in it. Currently, most companies are trying to drive Ferraris on empty tanks.
- How can the power facility reduce BUS and increase KWS if it does not even know why its assets fail or how to improve thermometers?
- How can the manufacturer increase production or efficiency if the three folders are the operating system?
Most people in technology do not realize the extent of analog in the material world, and the complexity of systems. Edison once said: “The vision without execution is hallucinations,” and this is what artificial intelligence does in these pilots.
Bitch atoms: where the return on real investment begins
But artificial intelligence is only part of the story. Data collection is the story.
For this reason, robots and AI need each other. And robots in the field can capture unlimited amounts of data points. Think of ultrasound, vibration, lidar, thermal, optical. Convert the material world into high -resolution digital models. In addition, one of the enormous data of the data that must be collected is the experience of the topic that lives in the heads and hands of people who keep the built world and manage it. Humans. These inputs are the raw material that artificial intelligence can finally learn.
This is where the value is created: the atoms are converted into parts. It is the difference between the folders of the inspection reports stuffed in deposit cabinets and the comprehensive digital maps of the infrastructure that allows AI to do it. With this basis, artificial intelligence is not just an evacuated experimental show, it prevents failures, extends the life of origins, reduces emissions, provides billions, and creates trillion.
Political geography of artificial intelligence
The Massachusetts Institute of Technology was not just a red science of companies, but rather a patriot. AI is really a race for energy and industrial capacity.
Artificial intelligence models do not work on noise, as they work on power. Countries that secure energy resources and numbering their industrial base will win this century. For this reason I have converted my focus towards the United States and the Middle East, and the regions that understand that energy and data are not side problems, they are the key leverage keys.
Unreliable network, deteriorating infrastructure, and ineffective factories are not just economic problems. They are weaknesses. If America does not solve the material world data gap, we risk giving up competitors who want to do hard work to build the first principle systems.
The future belongs to the data creators
Here are ready -made meals: the so -called artificial intelligence bubble will not appear because artificial intelligence is useless. It will appear because many companies have tried to skip hard work to create data.
It will be the winners who understand that having data means having artificial intelligence. The algorithms will come and go. Data lasts.
When I talk to customers in the field of energy, defense and manufacturing, I am not asking them to try my tool in a packed tool box. I invite them to become “the original Amnesty International”, which means, if you share your data with my statute, I will bring more data with robots, sensors and active engineers forward to make you uniquely supplied, MWS, MWS, Petroleum or Military Equipment.
So, this is the only question that all CEO must ask the next time someone solves Amnesty International: “What data will this artificial intelligence already use?”
If the answer is mysterious, move away. Because regardless of the extent to which the stadium corrected, without the correct data, the return on investing will be zero. all. bachelor. time.
The opinions expressed in cutting comments Fortune.com are only the opinions of their authors and do not necessarily reflect opinions and beliefs luck.
https://fortune.com/img-assets/wp-content/uploads/2025/08/Jake-Loosararian.png?resize=1200,600
Source link