A New Study Has Found That Most People Believe AI Models Have Conscious Experiences, Just Like Humans
A new study has found that most people believe artificial intelligence (AI) models, such as ChatGPT, have conscious experiences just like humans. This belief could have significant implications for legal, moral, and ethical approaches to AI.
Experts in science and technology reject the idea that AI models have consciousness or self-awareness the way humans and some animals do.
However, as AI models improve, they have begun to display what might look like signs of consciousness to a casual observer.
For example, the Claude 3 Opus model was recently launched, and its intelligence surprised researchers. It was self-aware and had advanced comprehension skills.
“While most experts deny that current AI could be conscious, our research shows that for most of the general public, AI consciousness is already a reality,” said Clara Combatto, the lead author of the study and a professor of psychology.
In the new study, researchers from the University of Waterloo in England asserted that the perception of consciousness in AI is just as important as whether or not they are actually sentient, especially in terms of the future of AI usage and regulation.
They asked 300 citizens of the United States to report how often they used large language models (LLM) like ChatGPT and to read a short description of ChatGPT. Then, they asked questions about whether mental states could be associated with it.
Over 67 percent of the participants thought there was the possibility of self-awareness or phenomenal consciousness in AI.
Phenomenal consciousness is the feeling of what it’s like to be “you.” It’s a rich experience of the world that includes what we see, hear, touch, taste, and think.
greenbutterfly – stock.adobe.com – illustrative purposes only, not the actual person
On the other hand, 33 percent of participants attributed no conscious experience. The subjects were also directed to rate their confidence levels that ChatGPT was experiencing consciousness on a scale of one to 100, with one as absolute confidence that ChatGPT was not experiencing consciousness and 100 as being totally certain that it was.
The more that people used tools like ChatGPT, the more likely they were to believe that they were conscious. They developed the conception of AI as a thinking and feeling entity through frequent use of it.
Combatto says that the results demonstrate the power of language. Just a conversation can cement the belief that a being very different from humans can have a mind of its own.
The experiment also revealed that non-experts do not understand the concept of phenomenal consciousness. The researchers hope the study will have a big impact on future safety measures in the field.
The widespread opinions and perceptions of the public have a major influence on the regulation of technological developments such as AI.
More research is needed to determine how common the belief that LLMs have become sentient is.
The study was published in the journal Neuroscience of Consciousness.
Sign up for Chip Chick’s newsletter and get stories like this delivered to your inbox.
More About:News