Research Shows While Most Americans Are Aware of AI, Few Pretend to Know How It Works — and Three-Quarters Wouldn’t Even Let It Pick Out What They Wear to Work
People may be fascinated with AI in the news, but there’s a long way to go before they trust it.
Purpose of this Study
Except for those hibernating in a cave, the winter and spring of 2022–2023 are nearly synonymous with “breaking news” about artificial intelligence or AI. It’s not that AI is exactly a new concept. We’ve been talking about AI since a computer scientist (John McCarthy) first used the term in 1956 while proposing a research conference at Dartmouth College. However, though the news stories are making everyday people more aware that AI is no longer the stuff of science fiction, the reality is that it is already integrated into both commercial software applications and operationalized into internal business processes.
Simply put, Artificial intelligence (AI) refers to computer systems that can perform tasks that normally require human intelligence, such as learning, problem-solving, and decision-making. And it’s not a fantasy. With the first public prototype of ChatGPT in November 2022 and subsequent more powerful and refined versions — not to mention the nearly overnight incorporation of this and other AI bot technology into search engines, commercial software and more — tangible AI is already here.
But are we humans ready to trust these programs and self-learning algorithms with the types of decisions that can truly alter the shape of our lives? Last month, we conducted our 2023 AI Trust Study to find out. We surveyed 1,000 U.S. adults about their awareness of AI, how it will impact their lives, and how much trust (or lack thereof) they have in its abilities.
Key Findings
- Most Americans are aware of the existence of AI, but the number who consider themselves familiar with how the technology works varies significantly by generation.
- Most respondents feel that AI will eventually affect their jobs, though they are divided on whether that impact will be positive or negative.
- When asked what AI should never do, Americans are least likely to want AI to make highly consequential life-and-death decisions.
- When presented with many other scenarios that directly or indirectly affect them, Americans still trust humans over AI by a wide margin.
- Men and women mostly agree on what they want or do not want AI to do, but men seem more open to letting AI handle a broader range of tasks.
Most Americans are aware of the existence of AI, but the number who consider themselves familiar with how the technology works varies significantly by generation.
To the average person on the street, AI — the real thing, not from Sci-fi movies — is still in its infancy. Yet, less than six months after the high-profile public launch of ChatGPT, the number of persons that consider themselves in the know is surprisingly high.
- 58% of respondents know that AI exists but don’t consider themselves familiar enough to understand how it works.
- Younger Americans are more familiar with AI than older generations. Nearly two-thirds (61%) of GenZ and over half (56%) of Millennials are aware of how AI works compared to only one-in-four (24%) of GenX+ who said the same.
Most respondents feel that AI will eventually affect their jobs, though they are divided on whether that impact will be positive or negative.
While 54% of Americans answered “yes” when asked if AI would impact their job somehow, those in a management capacity are more apt to believe the impact will be beneficial than rank-and-file workers. The latter tend to be negative or neutral about the effect of AI on their jobs, as do those with less overall awareness of AI.
- 54% of managers and senior decision-makers feel optimistic about the potential for AI to improve their jobs. Only 11% of this group think AI will negatively impact their jobs.
- Millennial managers are the most optimistic, with 73% of millennial managers responding that AI will have a slightly or very positive impact on their jobs, compared to 41% of GenX+ managers.
- However, GenX+ managers are more than three times more likely to think AI will negatively impact their jobs than Millennial managers (18% vs. 5.4%, respectively).
- Rank-and-file employees are less optimistic about AI. They are nearly twice as likely to think AI will hurt their jobs compared to managers and senior decision-makers (11% & 19%, respectively).
- Many rank-and-file employees are neutral about AI. 47% of rank-and-file employees think AI won’t affect their jobs (positively or negatively) compared to 32% of managers & sr. Decision-makers.
- GenX+ rank-and-file are the least concerned among the studied generations. 55% don’t think AI will affect their jobs compared to only 40% of GenZers and Millennials. This lower level of concern could be because GenX+ers are nearing retirement age and feel the major changes from AI will likely occur after they’ve already retired.
- Lower awareness of AI corresponded with greater concern about its impact on one’s job. Conversely, those who know more about AI are more positive about its potential impact on their job.
- GenX+ rank-and-file are the least concerned among the studied generations. 55% don’t think AI will affect their jobs compared to only 40% of GenZers and Millennials. (This could be because GenX+ers are nearing retirement age and feel the major changes from AI will be in the future after they’ve already retired.)
When asked what AI should never do, Americans are least likely to want AI to make highly consequential life-and-death decisions.
As fascinated or even awed as they are by technologies like the highly publicized ChatGPT, respondents draw the line when it comes to letting AI make critical, consequential decisions. These include situations where lives, safety, and even freedoms are at stake (those of their own or others). In other words, when asked if AI should ever be able to make decisions in these “life-and-death” scenarios, the answer was a resounding “no.
- 67% wouldn’t want AI to make life-or-death decisions in combat.
- 64% wouldn’t want an AI juror at their trial if charged with a crime.
- 57% would never want AI to fly an airliner (including taxi, takeoff, and landing).
When presented with scenarios that directly or indirectly affect them, Americans still trust humans over AI by a wide margin.
It’s not hard to fathom that the infancy of applying AI to real-life situations probably explains the reluctance to allow AI to make life-and-death decisions. However, we also asked respondents about a larger group of tasks that directly or indirectly affect them to see what Americans would be willing to allow AI to do. We found that even those who would consider letting AI do specific tasks still believe humans — either themselves or others — would perform the tasks better than AI:
From something as simple as picking a daily outfit to teaching school to writing laws—not to mention doing their own jobs—Americans aren’t yet willing to allow AI to make decisions or work tasks where the outcome will potentially affect them.
Men and women mostly agree on what they want or do not want AI to do, but men seem more open to letting AI handle a broader range of tasks.
As you can see, whether life-threatening or not, most respondent answers reveal that Americans overall don’t trust AI to do tasks and make decisions that would remove their human control over material yet everyday situations. In most scenarios, men and women align with what they would and would not want AI to do. Yet, in some cases, men tended to be more open than women to AI performing a wider range of tasks.
- 58% of men would be open to AI administering medicine in a hospital compared to 45% of women.
- 41% of men would be open to AI being a juror at their trial compared to 32% of women.
- 66% of men would let AI determine their salary/raise compared to 57% of women.
- 72% of men would be ok with AI checking the results of a mammogram or colonoscopy compared to 60% of women.
- 57% of men would be ok with AI doing their job compared to 48% of women.
Despite growing awareness of AI, the lack of understanding of how AI works appears to correspond with a surprisingly neutral attitude toward the technology and a lack of trust in its abilities—so far.
The study shows that most Americans are aware of AI, especially among members of more youthful generations. Although the respondents weren’t asked directly about the source of that awareness, some must partially be due to the recent swell of news coverage of new AI-driven chatbots like ChatGPT and Bard. Not to mention the application of these tools to search engines (Bing), word processors (Microsoft Word), voice assistants (Alexa, Google, and Siri), and almost every other software in between. And while only one in four GenX+ respondents know how AI works, most GenZ (61%) do understand this (or believe they do).
Why, then, is there such a reluctance for Americans to place their trust in AI to perform what might be considered basic tasks? It should not be a shock that respondents aren’t ready to accept an AI member in their jury trial, in the cockpit of their transcontinental flight, or in other positions where their life or freedoms hang in the balance, but to write fair and impartial laws? Not even to properly select what color shirt or blouse to wear to the office today?
Perhaps this pairing of broad awareness yet seeming wariness explains one of the study’s final findings: that Americans’ overall attitude towards AI is Neutral. That is, even though AI is here to stay, as seems likely, the jury is still out on when the populace will be ready to accept it, trust it, and relinquish control of what, until now, have been uniquely tasks of humans. It will take time and positive experience to build that trust.
John Michelsen, CPO and co-founder of Krista Software, said the survey’s results underscore that not only public awareness of AI but also making transparent how it works is critical to earning the trust of the people whose daily lives, jobs, and even freedom will be affected by the adoption of AI.
“When it comes to AI, trust is something that must be earned, and part of that involves transparency and understanding,” said Michelsen. “To accept that a faceless algorithm is acting in users’ best interest requires a level of understanding on both sides. When we responsibly integrate AI into our productivity apps, business processes, customer service centers and countless other applications, we can’t assume users and consumers will have a background in computer or data science. The AI-enabled application must be understandable and speak the users’ language if those users are to gain an appreciation, confidence, and trust in these systems.”
Methodology
Krista Software used the third-party survey platform Pollfish to conduct an online AI awareness and trust survey of 1,000 U.S. adults on Apr. 25-26, 2023. Researchers reviewed all responses for quality control.
