Anthropic has released its report, What 81,000 people want from AI. Anthropic conducted the research through their AI interviewer built into Claude, asking users to share how they use AI, what they hope it could make possible, as well as their fears about AI’s impact on them and the world. 

The research spoke with users from 159 countries across the globe, including 3,219 respondents in the UK, the AI firm claiming the survey was the “largest and most multilingual qualitative study ever” – at least on the subject of AI use. 

To conduct the interview, Anthropic posed a set template of questions to respondents, with follow-ups tailored by Claude based on respondents’ unique answers. 

What People Want From AI

So what exactly are people hoping AI will deliver?

The highest demand was for ‘Professional Excellence’ at 18.8%. This category grouped respondents’ desire for workplace support automation and enabled professionals to focus on strategic or higher value tasks. 

The second most common thing people wanted from AI was ‘Personal transformation’ (13.7%), closely followed by ‘Life management’ (13.5%). ‘Personal transformation’ encompassed cognitive wellbeing, mental and physical health as well as romantic connections in a small number. 

When Claude dug into respondents’ answers, often what was clear was participants’ desire for a better quality of life, supported by AI, allowing them to free up time to spend on higher priority things such as personal relationships and leisure activities. 

Respondents seeking “Life management support” valued help with removing mundane logistical and administrative tasks, enabling them to be organised and focused.

Other priorities emerged, though less prevalent, which included ‘Societal transformation’ where respondents hoped AI could solve big problems such as curing or detecting disease, and ‘Financial independence’ and ‘Entrepreneurship’ wherein people hoped for personal financial growth.  

Are People Getting What They Want?

About four in five (81%) respondents believe AI has taken a step towards their stated vision.

‘Productivity’ (32%) was the area respondents felt AI has delivered most value for them. Others felt factors such as ‘Cognitive partnership’ (17.2%), ‘Learning’ (9.9%), and ‘Technical Accessibility’ (8.7%) were all aided by AI. 

Whilst 18.9% of respondents didn’t feel AI had taken steps to deliver their vision, however Anthropic suggested, this percentage will continue to decrease as AI develops. 

Some of the success stories ranged from speeding up coding processes, to having more time to spend with family, to creating businesses, to overcoming speech and communication barriers. 

As well as the skills and time gains delivered, Anthropic suggested patience, availability and absence of judgement as key reasons respondents used AI. 

In extreme circumstances, some users (6%) reported using AI as their support system. However, some experts have warned against this, speaking to the risks and limitations of relying on AI for this type of support. 

Anthropic suggested that people’s vision for AI could broadly be categorised as a desire for more time, autonomy, and personal connection, however concerns were more varied – and worrying.

What People are Concerned About

The top reasons people were concerned about AI  included ‘Unreliability’ (26.7%), ‘Jobs & economy’ (22.3%), ‘Autonomy & agency’ (21.9%), ‘Cognitive atrophy’ (16.3%), and ‘Governance’ (14.7%). 

Only around 11% of respondents said they had no concerns or felt neutral about the technology, suggesting they felt confident that any problems that arose would be rectified. On average respondents cited 2.3 different concerns. 

As well as key concerns such as the job market and loss of cognitive ability, other significantly concerning issues were raised, such as bias and discrimination, political integrity and geopolitical concerns. 

Balancing Tensions

What the report found was that in most cases there are potential benefits and harms for each AI use case. For example, whilst learning from AI can unlock opportunities, becoming over reliant may lead to cognitive atrophy. 

Similarly, time-saving was one of the most desirable benefits cited, yet 19% of respondents were concerned about losing time from having to verify information from AI.

Sixteen per cent of respondents were hopeful AI could deliver emotional support, whilst 12% raised this emotional dependence as cause for concern.

Interestingly, the individuals who report benefiting from a certain feature are also more likely to be concerned about its use. In cases where someone valued emotional support from AI, they were three times more likely to fear becoming dependent upon it. 

Recommended reading

• Report: Appetite for AI Investment Higher Than Manageable for CIOs
• Scotland Publishes 5-year AI Strategy to Compete Globally
• How Worried Should We Be About Hostile State AI Collaboration?
• Scotland Leads UK in AI Productivity Gains
• Why 75% of AI Pilots Never Scale

Anthropic suggested that whilst concerns are valid, the benefits people discussed tend to refer to lived experience, whilst concerns were rooted in hypotheses. 

For each perceived benefit there were associated harms – and vice versa. 

Geographical Differences

The overall perceptions across the globe were mapped in the report, with 67% of respondents expressing net positive sentiment towards AI. In emerging economies, responses were overwhelmingly positive towards AI. Anthropic suggested this may be because early adopters in these countries view the new technology as a “ladder up rather than a threat”. 

The UK came out more sceptical than other regions, with a 63% net positive sentiment towards AI. Similarly, Germany scored a 64% positive sentiment average, France 65% and Ireland 65%, suggesting European sentiment is fairly consistent, and slightly behind other nations.