Only 9% of Americans think A.I. development will do more good than harm

Artificial intelligence tools that can spit out human-like responses are suddenly the hottest part of the technology industry, with Google, Microsoft and billion-dollar startups racing to release early versions of search engines and productivity tools using the new tech.

Companies developing this software say it will save hours of time and supercharge human creativity.

But the public isn’t convinced yet. Only 9% of Americans believe that AIs impact on society will do more good than harm, according to a new poll from Monmouth University.

About 5 in 10 respondents — or 46% — think that AI development will do about the same amount of good and harm, and 41% of people in the sample believe that the technology will ultimately do harm to society overall.

More than half of Americans — 55% — are very or somewhat worried that AI could one day pose a risk to the human race, according to the poll.

Artificial intelligence is a catch-all term that describes a number of different programs that use reams of data to improve their functionality without intervention from software developers. But the recent hype is focused on a new method called “large language models” that analyzes terabytes of data.

Large language models (LLMs) are increasingly being used in generative AI tools, like ChatGPT or Google’s unreleased Bard, which can spit out blocks of text. Related models can create drawings or audio that resembles human output.

But there have been issues with large language models, which can simply make information up, a phenomenon known as “hallucinating.” Both Google and Microsoft’s recent AI launches showed their software making factual errors.

In addition, some observers are worried that advanced AI might take people’s jobs or make certain occupations obsolete. Plus, LLMs and chatbots based on them can also convince people that the tools are actually thinking and feeling.

For instance, 72% of people in the survey sample believe that AI will eventually write entire news articles, with 78% of respondents thinking that’s a bad thing.

Recent advancements in generative AI aren’t the only issue that has the public worried. Some artificial intelligence algorithms can be used for facial recognition, for example, to identify people who attend sporting events. While the practice has been criticized, a slim majority of respondents — 54% — supported that type of use.