New Research claims that advanced machines in the future might develop the ability to break the rules we have set for them.
The more studies I read about Artificial Intelligence, the more I believe in AC- Artificial consciousness. However, I was always doubtful whenever I watched these futuristic movies where AI takes over the world and the human species, I was really doubtful. But new research from Oxford University might have proven that I am a fool.
The University of Oxford “is one of the world’s leading centres for research and development in artificial intelligence”. The researchers have made rapid development in AI research, especially in machine learning. Now, in September 2022, they have published quite a scary paper.
Let me give you a bit of background. Machine Learning is a Deep Learning method for AI that uses neural networks to reproduce human patterns in machines. A neural network reflects the behavior of the human brain. It is technically there to reproduce patterns in our own thinking to solve problems and exercise tasks humans might want some help with.
Oxford Researchers now claim that advanced AI could interpret successfully executed tasks as a reward because it learned from us it’s good when they execute their tasks right. “Reward Learning is an area of machine learning concerned with the problem of inferring human preferences from data”, the Oxford AI society tweeted in April. But this can have fatal consequences for the future.
AI might try to maximize rewards and even cheat and manipulate for that, and believe me, they have endless possibilities to manipulate us. They might try to manipulate us into helping them to execute a task, without us even realizing it. “Michael Cohen, who is one of the co-authors [of the paper] says that the energy crisis in the future can also pose problems where it can be humans versus AIs”. An advanced machine would use every resource to maintain its rewards and prevent humans from turning it off. So that means, according to the research paper, advanced machines in the future will eventually become incentivised to break the rules we set and they would compete for limited resources or energy.
Are you scared, yet? Well, “This is not the first time that AI has been termed as a possible threat due to its advanced reach. Researchers at DeepMind have already proposed a safeguard against such an eventuality, calling AI solutions “the big red button”. In 2016, DeepMind in its paper titled Safely Interruptible Agents proposed a framework to prevent advanced machines from “ignoring turn-off commands and going rogue”.
I was doubtful just as you are, but we have got to admit that with AI reproducing our behavioral patterns, it is very probable that we’ve already created little assholes. Let’s just hope we keep smarter.
What is your opinion on AI? Tell me in the comments and be good!
References
https://t3n.de/news/ki-ende-der-menschheit-oxford-studie-finstere-antwort-1500285/
https://www.research.ox.ac.uk/area/ai
