.webp)
1. What does energy greening have to do with AI?
“AI is very energy intensive. A tool like Chat GPT should 'train' you to learn information. Think of it like a doctor with 30 years of experience. You can ask them a question that you will get an answer to, but that was preceded by 30 years of training. AI training is based on the same principle, but in a shorter period of time. All this data processing requires an enormous amount of power, especially on a large scale.
Because it uses a lot of power, it has a big impact. Normally, all this generated energy is discarded. We ensure that the generated energy is not wasted, but is reused to replace fossil fuels when heating buildings. We run on green energy and use it twice!”
“AI shows existing ideas; it can't come up with new ones on its own yet. At least not right now.”
2. Aren't there easier ways to live more sustainably?
“Perhaps this comparison clarifies it: Chat GPT runs on 10,000 processors. The emissions of one such processor in a data center in a year are comparable to driving 10,000 kilometers.
For most people, it won't be the most obvious way to make a positive impact, but this is something we've encountered in our field. Essentially, training AI is like driving an SUV on the digital highway. A lot of heat is generated in a data center, after which air conditioning is used to cool it down. As a thermodynamist, I study energy changes, in particular the conversion of heat into energy. That's why early on in this process, I thought: why is this heat being wasted? We then discovered an entire world that we didn't know about before.”
3. What does AI mean to us in the short term?
“AI can do great things for us, but it will also have a major impact on certain industries. In the short term, we will become more productive, and some jobs will disappear. AI is ideally suited to handle help desk and call center work: answering 300 identical questions 8 hours a day is not a task you want to leave to the human brain. Of course, like all changes, this change raises resistance. But ultimately, we are one step closer to a better quality of life for everyone.”
“I predict AI fear of failure: stop writing texts yourself, because you think AI can do it better. The social impact will be greater than in any other area.”
4. How do you envision the future?
“So far, AI has failed to match human quality and there are still significant errors in the results. However, when Artificial General Intelligence (AGI) is fully developed, a computer can work like a human being does. That means it can do something it has never done before, without training or with very limited instructions. There will come a time when computers understand how human brains work and can mimic it.
Whether that hurts? Yes, definitely. When I write a text, I already catch myself thinking: am I going to write this myself or am I going to let AI do it? It's a kind of AI fear of failure, as it were. That will be even more difficult for teenagers: “I will never be able to write this paper as well as AI can, so I'm not even going to try it.” The social impact will be greater than in any other area. But for now, we have to remember that AI can only recreate existing ideas; it can't produce new ones yet. Not yet, at least.”
Curious about Leafcloud's work? Check out their website https://www.leaf.cloud/