Simone Zanetti AI’s Post

Transcript

Hey everyone, it's November 2024 and this year alone I've been doing over 40 masterclasses in artificial intelligence and seminars with large enterprises and governments. And one of the topics that comes out the most is of course copyrights and ethics, right? So what is the responsible use of AI? And these things are particularly hot for the fact that there aren't laws that regulate AI. Until now, there's been a small law that came out this year in May in Italy. And then of course, the EU AI act that will be active from December that states that all companies in Europe and all the companies that want to trade with Europe, will be forced to train all their employees in AI, specifically in safety, security and privacy. In fact, the privacy is one of the main topics. And because when you process the data with AI, you need to know where this data gets processed and follow the laws of your country. One of the things in terms of ethics that is very important, is to disclose to people when you're using AI. This is what happens also with advertisement in the 80s when they said you need to disclose when you use advertisement. Same thing when you use AI, you need to make people aware that you use it to generate content or that you use it to process data. This also builds trust because remember that AI is just a tool. Is a tool like a calculator. As a large organization or a government, you should see AI as something that empowers people. Not that replaces people, it's like a powerful computer or calculator. If you ask an accountant to do your taxes without a calculator or without a computer, it's probably possible it's gonna take a long time and it might make mistakes. Instead, if you use a calculator, it's going to be easier. If you use a computer, even easier. If you use AI, even easier. It's just that the next step. So let's empower people in terms of AI instead of replacing people. And in order to do that, we need AI literacy. People need to be trained, this is one of the jobs that we do at the Zanetti AI Institute to train people in organisations and governments into embracing AI and get the most out of it. So instead of saying "we can fire a number of people", you can say "we can do so much more with the people we have". One word about copyright because it's very important. Most of the generative AI software, they say that the use that you make... (if you have a paid account) that the use that you make of the art that you generate or the images or the text etc. The use is up to you, so they delegate to you the responsibility of using it. So the way that you can think about it is like if you have an artist like a colleague, that is very good. So if this artist is very good at drawing Spiderman and it draws Spiderman with your logo to do advertisement for your company, is that legal? I don't think so. Imagine AI like a digital artist. You can ask AI to do anything. If you can use it or not, it follows the same rules and boundaries as the normal, "analog world", you wanna put it that way? And lastly, AI in your organisation needs to have clear boundaries. Every time we consult with large organisations or governments, we create a new figure inside their compliance department called an "AI Auditor". Every time that you implement a large system with AI, you typically have one system, especially if it's a generative AI that generates something. And then you have an antagonist AI. And the job of the antagonist AI (Generative Adversarial Networks (GANs)) is to check that the first AI is due to what is supposed to do. But what if the antagonist AI (GAN) is not doing its job? Then you have an AI auditor, one person at the beginning, and then a team, that checks the data. Everything that is done with AI in the company follows the rules and the compliance that you decide. I hope that these few points have been helpful for your ethics exercise and if you have any questions, reach out to me. Have a good one.