What happens when machine learning goes too far

There’s always some truth in fiction, and it’s time to stay one step ahead of sci-fi dystopias and identify the risks machine sentience may pose to humanity.

While people have long thought about the future of smart machines, these questions have become more pressing with the rise of artificial intelligence (AI) and machine learning. These machines are similar to human interaction: they can help solve problems, create content, and even hold conversations. For fans of science fiction and dystopian novels, a looming question may be about to arise. What if these machines develop consciousness?

Researchers published their study on the topic Journal of Social Computing December 31, 2023.

Although no quantifiable data is provided in the discussion of artificial perception (AS) in machines, there are many similarities between the development of human language and the factors required for machines to develop language in a meaningful way.

“Many people concerned about the possibility of the development of machine sentience worry about the ethics of our use of these machines, or whether machines acting as rational calculators will attack humans to ensure their own survival,” said author and researcher John Levi Martin. “. “We were concerned that they would fall into some form of self-alienation by transitioning to a particular form of language perception.”

The main features that make this transformation possible appear to be: unstructured deep learning such as neural networks (computer analysis of data and training examples to provide better feedback), interactions between humans and other machines, and various Action continues self-driven learning. Self-driving cars are one example. Many forms of artificial intelligence already tick these boxes, raising concerns about what the next step in its “evolution” might be.

This discussion points out that it is not enough to focus solely on the development of AS in machines, but rather raises the question of whether we are fully prepared for the emergence of a kind of consciousness in machines. Now that artificial intelligence can generate blog posts, diagnose diseases, create recipes, predict illnesses, or perfectly tell stories based on its inputs, it’s not too far away to imagine what it would be like to have a real connection with a machine that already understands its inputs. state of being. However, the researchers of this study warn that this is exactly when we need to be wary of the output we receive.

“Being a linguistic creature is more about strategic control of information and leads to a loss of integrity and completeness…that’s not something we want in a device that’s responsible for security,” Martin said. Since we’ve made artificial intelligence Intelligence is responsible for a lot of our information and essentially relies on it to learn a lot in the same way the human brain does, so it has become a dangerous game when entrusting it with so much important information. Reckless way.

Mimicking human reactions and strategically controlling information are two very different things. A “linguistic being” may have the ability to be duplicitous and calculating in their responses. One of the important factors is, when do we find ourselves being played by machines?

The future lies in the hands of computer scientists who will develop strategies or protocols to test the speech perception capabilities of machines. The ethics behind using machines that have developed a language form of sentience or “self” awareness are not yet fully established, but it is conceivable that it will become a hot topic in society. The relationship between self-actualized humans and sentient machines is certainly complex, and the uncharted territory of this kinship will certainly bring with it many concepts about ethics, morality, and the continued use of such “self-aware” technology.

Provided by Tsinghua University Press