When it comes to trusting artificial intelligence (AI), men, millennial, and Gen Z workers generally have more faith in the technology than women, Gen Xers, or Baby Boomers, according to the results of a survey of more than 2,000 US adults.

The survey, the second of its kind conducted eight months apart, was performed by The Harris Poll on behalf of MITRE Corp., a nonprofit research agency that manages research for US government agencies in the aviation, defense, healthcare, homeland security, and cybersecurity areas. The initial survey on AI trust took place just before the launch of OpenAI's ChatGPT last Nov. 30.

Most respondents expressed reservations about AI when applied to government benefits and healthcare, and the latest survey showed a notable decline in trust in the past year.

“Late last year into this year, there was overwhelming excitement about generative AI and what it can do,” said Rob Jekielek, Harris Poll’s managing director. “For much of 2023, there has been substantial discussion about the potential negative implications of AI and how that has been accelerated by generative AI. [There has also been] discussion around lack of, and need for, more regulation, which may have led to a decline in AI trust.”

Only 39% of survey respondents believe AI is safe and secure, down 9% from the November 2022 poll, and 78% worry AI can be used maliciously. The poll indicates more work needs to be done on AI assurance and government regulation.

Ozgur Eris, managing director of MITRE’s AI and Autonomy Innovation Center, said “AI assurance” refers to providing maximum value while protecting society from harm. 

“From our perspective, AI has to satisfy expectations for technical, data, and scientific integrity, and produce desired and reliably effective outcomes. But this alone does not provide AI assurance,” Eris said. “For AI to be assured, it also has to permit organizational oversight and be safe and secure. It should also empower humans, enhance their capabilities, and augment their ability to achieve their goals, which means being interpretable by and answerable to those it empowers.”

AI should protect individual privacy, address inequities that might result from its use, and work in humanity's best interests in ways that are consistent with human values, ethics, rights, and societal norms, Eris added. “Not assuring these AI capability needs is likely to result in negative impacts..., whereas assuring them is more likely to produce more trustworthy AI, and to humans being better positioned to calibrate their trust in useful technologies,” he said.

The survey also showed that more than half (52%) of respondents believe AI will replace their jobs; 80% worry about AI being used for cyberattacks; 78% fear it will be used for identity theft; and 74% are wary of it being used to create deceptive political ads.

Just 46% believe AI technologies are ready for mission-critical use for defense and national security, down 8% from last year.

Perhaps most striking in the results was the difference in trust in AI depending on the gender of the respondents and when they were born. While 51% of men, 57% of Gen Z, and 62% of millennials indicated they're more excited than concerned about AI, only 40% of women, 42% of Gen X, and 30% of Boomers felt the same.

A majority of Gen Zers and millennials indicated they're okay using AI, but Baby Boomers were 20% to 30% less comfortable using the tech; only 37% of them were sanguine about AI for facial recognition on phones and personal devices, and even fewer, 29%, were okay with AI being used for targeted advertising on social media.

Most Gen Zers (54%) and millennials (58%) are willing to use AI for everyday tasks, but a much lower percentage of Gen X (39%) and Boomers (30%) are willing to do so. That gap shows up in other areas, with 51% of Gen Z and millennials comfortable with autonomous rideshare vehicles, compared to only 32% of Gen X and 20% of boomers.

According to Jekielek, the split between genders reflects a trend of women being less comfortable with technology because they have less knowledge about it and don’t use it as much as men. That, he said, leads to “less trust in AI, and less willingness to use AI.

“Older generations trust AI less than younger generations. As with women, less knowledge, use, and comfort with technology overall among older generations is a pattern we typically see in other research as well,” Jekielek said. “Younger generations are far more likely to embrace AI and its capabilities.”

Forty-seven percent of Gen Zers and 45% of millennials are comfortable with government agencies using AI to make decisions that directly affect themselves or their community, compared with 34% of Gen Xers and 24% of millennials. Younger US adults have even stronger concerns about AI and want assurance and regulations: 78% of Gen Z and 82% of millennials support regulation, compared with 86% of Gen X and 90% of boomers. 

“The public having reservations in trusting AI isn’t surprising, given the potential impact to jobs and the news around nefarious hacks like deepfake photos and videos,” Douglas Robbins, MITRE’s vice president of engineering and prototyping. “The public has been engaging more directly this year with consumer AI products and thinking through potential implications for their own healthcare, entertainment, transportation, or work experiences.”

Another recent survey of nearly 54,000 workers in 46 countries and territories by PricewaterhouseCoopers (PwC) found a mix of enthusiasm and concern as genAI tools such as ChatGPT move into the workplace. PwC’s Global Workforce Hopes and Fears Survey found sizable pockets of the global workforce are eager to learn new skills and embrace AI.

One differentiator among respondents: those with specialized training expect to see more of an impact from AI on their career, in both positive and negative ways. However, those without specialized training are less likely to anticipate an impact from the technology; in fact, 22% of global respondents say they don’t think AI will impact their job at all.

As genAI applications such as ChatGPT and Bard have entered the workplace, more than half (52%) of PwC survey respondents found at least one positive statement about the impact of AI on their career, saying it will increase productivity, bring opportunities to learn new skills, or create job opportunities.

Many respondents chose at least one negative statement, saying it will require new skills they might not be able to learn (18%); will negatively change their current work (14%); or will replace them in their current roles (13%).

Just 36% of respondents agree that the skills needed to succeed in their job will change significantly over the next five years, and only 43% say they have a clear sense of how the skills required for their job will change.

"It’s worrying that the majority of workers do not appear to have clarity on how their job requirements may change," the PWC study said. "If employees don’t anticipate or understand this, they may not be adequately prepared to acquire the new skills necessary to remain relevant and effective in their roles."

A "deeper concern" involves less specialized workers who appear the least likely to see change coming; only 15% of those respondents said the skills required to do their jobs will change in the next five years.

"This could leave workers who lack specialized training particularly vulnerable to job losses as skills continue to evolve, and as companies augment (or replace) jobs with automation, AI, or both," the study concluded.

PwC recommended that companies ask whether their workforce has the skills necessary to transform.

"Every leadership team should be able to draw a direct line from the capabilities they need to grow and innovate to the specific business outcomes they want to achieve, including transformation. But this isn’t a static exercise. Leaders must also be prepared to adjust the plan — repeatedly — as the environment changes," the study said.