Researchers from the University of Bath in the United Kingdom and the Technical University of Darmstadt in Germany say that there is no need to be scared of AI as it has no skill or potential to think independently and act upon it. According to this research, all the AI models need some instructions to function and they are easily controllable and predictable. People are unnecessarily scared of AI or the harm it can cause to humanity but in truth, AI models are clueless without human instructions.
Some influential tech visionaries like Elon Musk and OpenAI’s chief former scientist Ilys Sutskever have pushed a concept about artificial general intelligence (AGI) but AI models do not have the ability to take over human intelligence. Yes, they can learn a new language and be proficient in it, but that can only be done after they are given instructions.
Researchers explain the concept about AGI that it would be capable of learning and understanding things just like a human being and can carry the same intellectual tasks as them. But the research says that current sets of LLMs do not possess this ability even if they are trained on large datasets. AI models can surely answer questions about social problems even if they are not trained to do so but that is because of in-context learning (ICL) which refers to completing tasks based on a few examples presented to AI models.
The researchers of the study said that this ongoing fear of AI models is not making people see the actual threats use of technology can possess. One of the threats is AI’s lack of reasoning and planning which can be disastrous. AI is also generating fake news, deep fakes and AI bots which can disrupt people’s thoughts and can be overall hazardous. AI also has a tendency to hallucinate which makes it harder for users to communicate with them.
Image: DIW-Aigen
Read next: Google Updates Its ‘Trending Now’ Tool To Provide More Insights On Search
