Above, we spoke briefly about the real and hypothetical risks of AI. Below, we describe each of them in detail. Real-life risks include aspects such as consumer privacy, legal issues, AI biases, and more. And the hypothetical problems of the future include aspects such as AI programmed to cause harm or the development of destructive behavior by AI.
There are a myriad of AI-related risks that we face in our lives today. Not all the risks of AI are as great and worrisome as those of killer robots or intelligent AI. Some of today's biggest risks include aspects such as consumer privacy, biased programming, danger to human beings, and unclear legal regulation. One of the main concerns cited by experts concerns the privacy, security and artificial intelligence of consumer data. Americans have a right to privacy, established in 1992 with the ratification of the International Covenant on Civil and Political Rights.
However, many companies are already circumventing data privacy violations with their collection and use practices, and experts are concerned that this situation may increase as we begin to use AI more. Hallucinations refer to the mistakes that AI models usually make because, although they are advanced, they are not yet human and rely on training and data to provide answers. If you've used an artificial intelligence chatbot, you've probably experienced these hallucinations because of a misunderstanding or a blatantly incorrect answer to your question. Like hallucinations, deepfakes can contribute to the mass dissemination of false content, leading to the dissemination of misinformation, which constitutes a serious social problem.
The advanced capabilities of generative artificial intelligence models, such as programming, can also fall into the wrong hands and cause cybersecurity problems. Litan says that, while vendors that offer generative artificial intelligence solutions often assure customers that their models are capable of refusing malicious cybersecurity requests, these providers do not provide end users with the ability to verify all the security measures that have been implemented. How can artificial intelligence be dangerous? In addition to the concern that autonomous weapons may acquire a “mind of their own”, a more imminent concern is the dangers that autonomous weapons may pose to a person or a government that does not value human life. Once deployed, they are likely to be difficult to dismantle or combat. Invasion of privacy and social qualification.
One of the most pressing dangers of AI is technosolutionism, that is, the view that AI can be considered a panacea when it is nothing more than a tool.3 As we see more advances in AI, the temptation to apply AI decision-making to all social problems increases. However, technology often creates bigger problems in the process of solve smaller problems. For example, systems that simplify and automate the application of social services can quickly become rigid and deny access to migrants or others who go unnoticed, 4.In addition, it's only a matter of time before the first autonomous drone with facial recognition and a 3D-printed rifle, gun, or other weapon becomes available. Watch this video from Slaughterbots to get an idea of this.
Artificial intelligence systems that create false content also carry the risk of manipulation and conditioning by companies and governments.
