Kate Alessi
Vice President and Managing Director, Google UK & Ireland
For 25 years, the UK has been one of Google’s most important homes, and today it sits at the centre of our work on AI. Google DeepMind, founded in London by Sir Demis Hassabis, is a world-leading AI research lab committed to building AI responsibly, and creating products like the Gemini app and AlphaFold, which are bringing real benefits to society and delivering scientific breakthroughs.
The UK's world-leading research and science capabilities, alongside its talent and universities, make it a natural place for Google to build and grow. From our King’s Cross campus to our first British data centre at Waltham Cross, we continue to invest here for the long term.
My role is to ensure our technology enables people, businesses and the economy to grow. I’m proud of our impact. Public First’s analysis estimates that Google’s products, platforms and tools helped support £140 billion of economic activity across the UK in 2025, up 19% from £118 billion in 2023.
That is equivalent to the economy of Greater Manchester. Sixty billion pounds (43%) of this is accounted for by British small and medium-sized businesses using Google products to innovate and grow, while Google tools also help support £23 billion of UK exports.
An increasing share of Google’s value to the UK now comes from our newest and most important innovation: Gemini. Gemini is helping British people be more productive in their work and their personal lives, not only saving them significant time and helping them manage day-to-day priorities, but also helping them upskill.
When I started in this role, one of my first priorities was to ensure Google helps people to make the most of technology in their lives. We know from previous research that more than three quarters of adults in Britain feel “stuck” relative to where they want to be in their careers. They feel held back not by ambition, but by gaps in confidence, networks and skills. Used well, AI can help close those gaps and bring new opportunities for personal progression within reach.
This report unpacks that potential. The research found that even after accounting for differences in age, sector, gender, ethnicity, education and business size, AI use is associated with greater opportunity: more time saved, new skills gained and a higher chance of progressing at work. It is one of the few levers we have that can genuinely raise the floor for all.
But these rewards are not yet being felt by everyone. To fix that, the Government, employers, tech companies, unions and educators all need to meet the challenge of equipping Britain with the skills to use AI effectively. When more people can use AI well, everyone can win.
Our job is to help make the opportunity of AI available to everyone. This report is part of that effort: combining new evidence on Google’s contribution to the UK with one of the most comprehensive studies of AI adoption in Britain. It also contains practical recommendations for how to level up AI use, and a new AI Skills Quiz, designed to help people map and grow their AI skills.
Google has been part of British life for over a quarter of a century. Since opening our UK office in 2000, we have helped people across the country find information, build businesses, learn new skills and get more done — from Search and Maps to Gmail, Android, Workspace, Cloud and now AI with Gemini.
Where Google’s impact lands
Our products and platforms today generate an estimated £140 billion of economic activity in the UK every year, up from £118 billion in 2023.
Small and medium-sized businesses account for £60 billion of that — or 43% of the total — using our tools to compete, grow and reach new customers.
Google Search and Ads also support over £23 billion of UK exports, helping British businesses sell to customers around the world.
And people also benefit directly. We estimate that through time saved, better choices and easier access to information, Google’s free services deliver £465 in consumer benefit per person each year — roughly equivalent to a monthly food shop for a typical family in the UK.
Google is also helping to improve delivery of Britain’s public services. A recent study jointly conducted in partnership with the NHS and Imperial College showed how AI software developed by Google Health could slash clinical workloads while driving standards up.
Helping people get on
Beyond the headline numbers, Google tools are also helping people advance at work and in their careers.
An average worker now saves 1.5 hours per week using Google products — the equivalent of a £42 billion boost to national productivity, or roughly twice the size of annual public sector R&D spending. AI is increasingly part of this story. Over 20 million UK workers — more than two thirds of the total — have used Google tools to build their AI skills. Among Gemini users, 87% find it helpful, 70% say it saves them significant time, and nearly half use it to learn how to use AI better.
For younger workers, those tools are also part of career progression. Nearly three quarters of Britons under 25 have recently used a Google tool to help them apply for a new job, and most have also used one to boost their job-related skills.
These young people already use AI tools at work for everything from straightforward tasks like managing calendars and writing emails to more complex ones like data analysis and financial calculations. 57% say they use a Google platform at least once a day.
Google’s commitment to the UK
AI could unlock enormous value for the UK economy. But capturing it will depend on whether people across the country have the skills, access and confidence to use AI well.
Google has been investing in this challenge for over a decade. Since 2015, Google Digital Garage has trained over 1.2 million people in the UK with free digital and AI skills.
This year we launched AI Works for Britain, a multimillion-pound programme delivering practical AI training to high streets, universities and Jobcentres via pop-ups and a Department for Work and Pensions partnership. Additionally, Google.org is funding The Good Things Foundation to equip digitally excluded communities with AI skills for the future of work. Google.org has also provided funding to INCO and Chance, enabling participating universities to deliver practical AI training and career coaching. This is helping students transition into sectors that INCO research shows AI is transforming.
After accounting for differences in age, sector, education, gender, ethnicity and the size of business you work in, we find that people who use AI well are more likely to get promotions, pay rises, and progress faster in their careers.
We are witnessing a rare moment in history where the speed of a technology’s adoption is almost as remarkable as the scale of its promise. AI is set to fundamentally reshape how humanity works, creates, and solves problems — and the British public isn’t waiting around for it. Last year, only 34% of the British workforce were using AI. Today, that number stands at 73%.
While the rate of AI adoption is remarkable, the depth of that adoption is where the challenge lies. Though many are now using AI on a regular basis, most people are only scratching the surface of what AI could enable them to do. Fortunately, our research shows that many of the behaviours and skills required to harness AI’s full capacity to unlock opportunities and personal progression are straightforward to learn.
But upskilling requires effort, and not just from the end user. More needs to be done by business leaders and policymakers to equip people with the confidence, capabilities, permission and habits to use AI well. Google has been investing in this challenge for over a decade. Since 2015, Google Digital Garage has trained over 1.2 million people in the UK with free digital and AI skills.
This year, they launched AI Works for Britain, a multimillion-pound programme bringing practical AI training to high streets, universities and Jobcentres — through pop-ups and partnerships with the Department for Work and Pensions and the Good Things Foundation.
In this research, we have explored not just who is using AI in the UK, but how they are using it, what separates basic use from effective use, and what it takes to build the skills that make the difference. These findings provide a practical guide for government, businesses and individuals to unlock AI’s benefits, using the technology to help people advance in ways that would otherwise be out of reach.
Increasing AI skills will also help address long-standing inequalities — across genders, regions, educational backgrounds — but this will only happen if the Government and businesses act to equip everyone with the skills and tools to benefit.
For individuals, our evidence suggests that those who use AI most extensively in their daily lives are already seeing the greatest rewards.
Have not yet built AI into their daily lives. As a result, they report almost no time saved either at work or in their personal lives.
Use AI occasionally, mainly for quick questions and simple drafting tasks. Even at this early stage, they report saving 2.7 hours a week across work and personal life.
Use AI more regularly for tasks such as drafting, research and problem-solving. They save 3.7 hours a week on average, and are more likely than Beginners or Non-users to say they have been promoted in the last year.
Use AI in deeper and more creative ways, including automating tasks and experimenting with agents. They save 7.7 hours a week on average, and are the most likely group to report promotions, higher average salaries and greater confidence at work.
Anyone can use AI to get ahead
Even when controlling for a person’s background, age, education, job, gender, ethnicity and the size of business they work in, advanced AI users remain significantly more likely to report promotions, pay rises and faster career progression — suggesting the gains track with how people use AI, not just who uses it.
Right now, the UK’s most adept AI users are more likely to be younger, university graduates, male, London-based and working in knowledge-intensive sectors such as information and communications. But these patterns are not fixed, and a little support can have a big impact. With more engagement and targeted help, far more people could reach similar levels of AI proficiency.
Advanced AI users are not securing greater productivity and progression because of deep technical expertise. Instead, they have adopted simple, learnable habits: asking better questions, refining prompts, using AI to critique their own work, applying it to recurring tasks and sharing useful examples with friends, family and colleagues. Anyone can become an Advanced user, working with AI to better achieve their personal and professional goals.
Creating the conditions for productive AI use at work and home
AI adoption is booming, but for a large chunk of the population their use of AI is relatively basic. Our research shows only 15% of the UK are engaging with AI in a way that unlocks significant benefits.
The UK’s challenge is the depth of that AI adoption, and it’s one that can be solved with determined and focused action. In this report, we set out actions that individuals, employers and government can take to ensure everyone in society has the skills and confidence to capture the full upside of AI.
A practical toolkit to boost Britain’s AI benefits
Alongside this research, we developed a new AI Skills Quiz, that allows people to better understand their AI usage, see how they compare to the nation, and get practical recommendations to unlock further benefits from AI tools: from trying more useful prompts to reusing workflows, uploading files or asking AI to critique work.
Our research identifies the habits, behaviours and workplace conditions most strongly associated with effective AI use, informing the practical recommendations for individuals, employers and policymakers set out below.
Practical actions to level up AI across Britain — for Government, employers and people
Five things people can do
The most immediate step-up comes when people move beyond just checking facts and start using AI as a partner to help with generating, writing, creating, analysing and refining ideas. Beyond this, Advanced users ask AI to challenge and improve outputs over multiple rounds and are comfortable comparing between different AI tools (such as Gemini and NotebookLM) — knowing when to switch modes from fast models to longer thinking ones.
Often the easiest way for people to improve their AI use is by asking the system to help them pose better questions. But only a third report ever having done this and among Beginners it is virtually unheard of. Doing so helps move users beyond vague requests to clearer instructions. Users can use AI to help improve prompts by explaining the task more specifically, providing relevant background information, suggesting the intended audience, and setting expectations about the format or style of the response.
People should be proactive in asking their employer about AI use policies, including which tools can be accessed and what information can be shared with them. Keeping up to date with workplace guidance helps people feel more confident about using AI in their work, while avoiding risks around privacy and security.
Adoption spreads when people see examples that are relevant to their life and work. Colleagues, friends and managers can make AI feel practical by showing what they use it for, what works well, and what to avoid.
For the most confident users, creating personal agents can help them delegate workflows that carry out multi-step tasks automatically — whether monitoring information (such as weather patterns), managing routine administration (such as calendar schedules), tracking opportunities (such as job openings), or helping coordinate projects (such as organising a family holiday).
Five things employers can do
Our data shows that employees want to experiment with AI, but uncertainty around security and compliance can hold them back. Leaders can address this by approving a set of trusted tools and modelling the behaviour they want to see — giving clear boundaries about appropriate use and emphasising that building AI confidence is often a process of trial and error.
Most employees don’t need a long legal document about AI use in the workplace, but want simple guidance on what is encouraged and what is restricted. Our research has shown that these kinds of policies can be a simple step towards significantly increasing AI use.
A large part of AI’s value lies in its ability to be applied across an exceptionally wide range of use cases. Yet this versatility can also be a barrier to adoption. With so many potential applications, many people struggle to identify where to start with AI, often because they don’t know what they don’t know. People want to learn how to use AI, with 63% of adults interested in receiving training to help them use AI more productively. Our experience of delivering AI training is that role specific hands-on training drives a significant boost. We found a 10:1 Return on Investment for AI training, significantly more than for other comparable skills programmes.
People are more likely to become Advanced AI users when they are encouraged and incentivised to experiment, learn from colleagues and receive support from managers. Employers should encourage employees to share what worked, and managers should provide visible support for those testing new tools and workflows. Employers could also create incentives for productive AI use by employees — such as internal awards, or factoring it into performance reviews.
Our data shows that employees are interested in using AI to improve their work, but not all businesses are set up to encourage use or get the most out of AI when it is deployed. Employers should build a plan, working with trusted providers to identify what changes they need to make to maximise the return on investment. This will range from training and permissions for employees to reviewing how data is stored and accessed, examining what functions can be automated and assessing the range of AI tools available. Without the right structures and encouragement in place, the benefits AI delivers can only go so far.
Five things Government can do
Currently, less than half of public sector workers have received training on how to use AI tools. It should be a priority to ensure tailored, work-relevant training is provided to everyone as a matter of course. Beyond this, senior public sector figures should show leadership by giving access to approved AI tools and offering clear use cases and permissions. This twin approach ensures public servants have the confidence, permission and practical support they need to use AI effectively.
Efforts should target areas and business sectors with large numbers of Non-users and Beginners in particular — the groups where better adoption would have the greatest impact and where targeted support can help ensure AI’s benefits are more widely shared.
The Government should take any opportunity it can — including through government-to-business communications — to signpost available AI adoption guidance and best practice guides developed by industry groups, arm’s-length-bodies and trade unions. The recently published AI Champions’ AI Adoption Plans, Skills England’s guidance, and practical examples of AI adoption from industry-led coalitions should be incorporated into the guidance available on Business.gov.uk.
The UK’s ambition to be a global AI leader depends on the British people having continued access to the best tools. The recently published UK AI Hardware Plan, and the AI Opportunities Action Plan before it, provide a firm foundation to build on. Other factors key to physically building out AI cannot be ignored either — such as access to reliable, affordable energy, particularly for the UK to remain competitive in attracting further AI infrastructure investment, alongside which the Government should strengthen other forms of UK AI infrastructure by opening up high-quality public datasets to make it easier for innovators to integrate trusted information into AI tools.
Government should continue to take a pro-innovation approach to AI regulation. The goal should be simple, businesses should feel able to test, experiment and adopt AI where it can improve how they work, while staying within clear guardrails. Different sectors face different risks, from finance to health to retail, so regulators should be encouraged to give practical guidance that is relevant to the day to day of doing business in different parts of the economy. Done well, regulation should not make businesses hesitate. It should give them the confidence to try new AI tools, learn what works, and scale successful new innovation.
AI has moved quickly from novelty to normality. 88% of the British population have now used an AI tool, rising to 96% among 18–24s.
Despite two thirds of the British population (65%) having only begun using AI in the past year, almost three quarters (73%) now use AI tools at least monthly at work. In March 2025, 12% of British adults said they use AI at work on a daily basis; that figure has now more than doubled, standing at just under a third (28%). For a sizable group of people, AI is becoming part of their everyday professional toolkit.
AI is also becoming a personal habit. Over half (57%) of adults use AI tools at least weekly in their personal lives, rising to 71% among 18–24s. People are using AI to answer questions, plan tasks, learn, create, compare options and solve everyday problems.
Gemini: Driving AI across Google’s services
Google’s Gemini helps people with everyday tasks. It’s increasingly being used across Google Search, Android, Workspace and other services — reflecting how the technology is improving the products people use every day.
More than four fifths (82%) of people report having used AI at work, of which 73% use it on an at least monthly basis; 59% use it on an at least weekly basis and 28% on a daily basis. People are using AI tools to help complete a range of tasks, including to answer simple questions (true for 70% of respondents), learning or problem solving (56%), creating or improving content (51%), research and analysis (49%), and role-playing important conversations (36%).
Tools enhanced by LLMs like Gemini also regularly spill over from the workplace and into everyday life. One quarter of adults report using AI daily in their personal lives, and nearly three fifths (57%) report using it on a weekly basis. People say they use AI in their personal lives for a range of tasks, such as dealing with practical problems, making better financial decisions, pursuing personal projects and staying organised.
Adoption is broad, but depth of use is uneven
High adoption and even high frequency can still hide shallow use.
For this report, we group the population into four types of AI users. Across these groups, the key difference is not just how often people use AI, but how savvy that use has become. The biggest gains appear when people move beyond basic tasks into higher-value habits.
More advanced use is associated with more time saved, higher reported monetary benefits, higher average salaries and stronger reported career outcomes.
Advanced users of AI reap the rewards
Who is benefiting most from AI?
Anyone can benefit from using AI, but certain groups are already leading the way, harnessing AI for their own personal progression.
Age
Age is the sharpest differentiator. Younger people are far likelier to be Advanced users than older generations, with users aged 25–34 making up 29% of Advanced and just 4% of Non-users; on the other hand, users aged 65+ make up 54% of Non-users and only 3% of Advanced.
AI confidence peaks in the 25–34 bracket, perhaps because people in this group are more likely to have experienced AI in the workplace — tallying with our analysis suggesting workplaces are key drivers of AI adoption and upskilling. Greater autonomy through seniority, alongside stronger incentives around career progression and skills development, likely encourages them to integrate AI into their daily workflows. Having spent longer in the workforce than younger adults, they may also be better placed to identify which tasks can be automated, accelerated or improved through AI.
Gender
Adoption trends also vary between genders. Women are overrepresented in lower-use groups: 59% of Non-users and 55% of Beginners are women, compared with 41% of Advanced users.
This difference in engagement with AI also manifests in differing perceptions. Women are less likely than men to believe that AI will have beneficial impacts for the economy, society, and for their own careers; unsurprising given that optimism towards AI seems to scale with how deeply someone uses the technology. Men are more likely than women to report having used AI to help them carry out a wider range of tasks. And previous research we conducted also showed that women are less likely to use AI in the workplace without being given explicit permission — which naturally acts as a brake on adoption.
How are women over 55 using AI?
Older women are one of the groups most at risk of missing out on AI's benefits. Only 4% of women aged 55+ are Advanced users, compared with 15% of the population as a whole. Just 2% say they know a lot about AI, and 22% say they would not know where to start.
But the gap is not driven by lack of interest. Across the UK workforce, interest in AI skills training has risen from 60% to 69%, and there is no evidence the appetite stops at any particular age (although men do report higher levels of interest than women, and this gap has not notably changed with time).
This suggests that the gap in capabilities is fixable. A 2025 analysis of Google’s AI Works pilots — which deliberately focused in part on the education sector, given the high proportion of women working in it — found that a few hours of practical training can quickly change behaviour. Among women aged 55+ in the pilot cohorts, daily AI use more than tripled from 9% to 29%, while weekly use rose from 17% to 56% after three months. The lesson is clear: with relevant examples, hands-on training and permission to experiment, groups who are furthest behind can catch up fast.
Education
Education is currently another dividing line, with graduates much more strongly represented among higher-use groups. Forty percent of Intermediate users and 60% of Advanced users are graduates, compared with 25% of Non-users and 26% of Beginners.
This does not mean, however, that AI is the exclusive preserve of younger, university-educated workers, or those who are particularly tech savvy. As discussed earlier, practising simple, easily-formed habits can quickly move someone up the AI skills ladder.
But it does mean that the current pattern of adoption risks reinforcing existing advantages. Advanced AI use is associated with greater opportunity — more time saved, new skills gained, and a higher chance of progressing at work. And these advantages persist even when we account for differences in age, occupation and socio-economic background.
By democratising understanding and high-level capabilities, AI has the potential to help overcome long-standing divides in access to opportunity. But we should be impatient for that change to occur. It is the job of governments, employers and educators to accelerate the uplift of AI by taking deliberate actions to equip people with the skills, confidence and permission to use AI well.
Which parts of the country are benefiting most from AI?
Advanced users are found across the country, but they are not evenly spread. London stands out with 22% of Advanced users based in the capital, compared with just 6% of Non-users. At a constituency level, the 10 constituencies with highest Advanced user density are all found in London.
Manchester, Leicester, Nottingham, Bristol and Cambridge also have a high density of Advanced users, reflecting the presence of large employers, universities, digital clusters and knowledge‑intensive roles outside the capital.
This mirrors a wider pattern in the research. The most capable AI users are more likely to work in knowledge‑intensive sectors, hold senior roles and have greater exposure to digital tools at work. The clear risk is that AI could deepen existing geographic advantages, unless workers across all regions have the same access to tools, training, permission and practical examples.
The first phase of the UK’s AI adoption journey is well underway. Adoption is no longer the main story — the issue is now whether people are using it well enough, often enough, and confidently enough to help them achieve their goals.
Which parts of the economy are benefiting most from AI?
AI is a general purpose technology with the potential to impact a variety of sectors across the UK's economy, and businesses in all sectors can benefit by adopting it and learning from those further along in their AI journey.
Adoption is advancing fastest in sectors with the greatest exposure to AI. Information and communications remains the clearest early adopter sector, with the strongest concentration of Advanced users, more digitally intensive roles, and more workers comfortable experimenting with new tools.
Elsewhere, progress is mixed. Some sectors are already putting the right foundations in place through clearer policies, visible permission to use AI, and better access to tools, training and peer learning. Others remain at an earlier stage, with no settled playbook for turning experimentation into productivity gains.
We also see differences opening up between the public and private sectors. In the private sector, some 30% of workers are using AI daily versus 24% of workers in the public sector. It is not just frequency of use either — private sector workers are more likely to be Advanced users than public sector workers (21% compared to 15%).
This is significant because the gap is not inevitable. By identifying which sectors are leading on adoption and which are building the conditions for more effective use, empirical lessons can be learnt to allow support to be targeted to where it is most needed, allowing more workers and businesses to move up the AI ladder.
Turning AI awareness into AI advantage
Google’s AI Works for Britain initiative reflects a growing recognition that widespread access to AI tools does not automatically translate into meaningful economic benefit. As our research shows, while AI adoption in the UK is already high, many people are still using these tools only in basic ways. AI Works for Britain is designed to put research into action and help people move beyond simple prompts towards higher-value uses of AI that can support career progression, confidence and productivity.
Our new nationwide AI upskilling initiative, AI Works for Britain, including nearly £2 million in Google.org grant funding, is being rolled out through the following ways:
Nationwide ‘Squeeze the Juice’ pop-up bars: Bringing practical AI training to communities across the UK through hands-on demonstrations that show people how to use AI to save time and solve everyday problems.
Gemini university tour and AI Skilling for Students: Equipping students with the AI skills, training and career support needed to succeed in an increasingly AI-enabled labour market.
Grassroots upskilling to ensure no one is left behind: Expanding access to AI skills through community organisations and Jobcentres to help people at risk of digital exclusion prepare for the future of work.
Supporting the future of healthcare
Healthcare systems around the world are under increasing pressure from ageing populations, rising demand and workforce shortages. AI has the potential to address some of these challenges by supporting clinicians with diagnosis, reducing administrative burdens, and enabling healthcare professionals to spend more time on direct patient care. Through Google Health and partnerships with researchers, hospitals and universities, Google has been exploring how AI can be applied responsibly to improve health outcomes and support healthcare delivery.
A recent study conducted by the NHS, Imperial College London and Google revealed AI’s potential to help strengthen breast cancer detection efforts. The experimental research AI system identified 25% of the “interval cancers” that were previously missed — the cases that typically slip through traditional screenings and only surface after symptoms appear, when they become more challenging to treat. The peer reviewed study also showed that AI reduced the time taken to read a scan by almost a third (32%).
Innovations such as these demonstrate AI’s potential to help a healthcare service stretched to its limit and give clinicians more time for patient care.
AI’s impact rises in line with capability. Advanced users save more time, report higher monetary benefits and are more likely to say AI is helping them at work. And the upside isn’t limited to the workplace. Our research suggests that the AI skills people develop at work carry over to their personal lives too.
Evidence that using AI effectively boosts career outcomes
When controlling for age, gender, ethnicity, sector, education and business size, Advanced users are more likely to report positive career outcomes. They are 84% more likely to have been promoted in the last year, 55% more likely to have received a pay rise, 157% more likely to say they are progressing in their career, 87% more likely to feel confident at work, and 88% more likely to report positive performance reviews.
Perhaps most importantly, more capable users are not only seeing immediate benefits but they also imagine them continuing, which may feed into higher reported ambition for the future. Advanced users are also significantly more likely to say AI can help them deal with tasks that many people dread, but are essential for personal progress.
Evidence of effective AI use benefiting personal life
The same pattern is playing out in people’s personal lives. Intermediate and Advanced users are more likely to say AI helps them learn complex skills more quickly and achieve goals that previously felt out of reach. The most advanced AI users on average save 7.7 hours per week across work and personal life — around 400 hours a year. Intermediate users report around 192 hours a year saved.
How AI is helping people professionally and personally, in their own words
[I used AI to] rehearse job interview questions.
I used AI to help recreate a family recipe that had never been written down properly.
AI helped reset my boiler after a powercut.
I use AI as a personal tutor that explains and simplifies almost any topic on demand.
AI helped me understand some tribunal terms that I was struggling to fully grasp as I did not have a lawyer, which helped me build my case and win.
[AI helped] develop a career pivot.
AI helped me compare prices and deals across multiple websites to find the cheapest option for a laptop before I bought it.
I used [AI] to plan my perfect wedding.
I use [AI] as a personal development coach.
I used AI to figure out how to fix a leaking tap myself instead of calling a plumber. It gave me step-by-step instructions, so I saved on repair costs.
Using AI to revolutionise transport
Autonomous vehicles (AVs) provide a useful illustration of how AI could reshape sectors far beyond technology itself. AV services are already operating commercially in cities around the world. Waymo, which began as a Google research project, is now running a commercial service in eleven US cities, offering over half a million trips per week, and has announced plans to deploy a service in London by the end of the year. This demonstrates the transformative effect AI could have on how people and goods move around cities and regions.
The longer-term opportunities could be significant. Safer roads, more equitable access to mobility, and improved efficiency in how roads and the kerbside are used. Autonomous vehicles will also create new jobs in the UK. Increasingly advanced vehicle technology will mandate high-skilled manufacturing roles. And, an entirely new suite of roles will be needed to support the behind-the-scenes maintenance and coordination of autonomous vehicle fleets.
The habits that make the difference between heavy and light AI users are neither specialist nor difficult to learn. Advanced use doesn’t require knowing how to code and build models, and most Advanced users in our research don’t possess deep technical expertise. Rather, they simply understand how to ask better questions, test answers, refine outputs, reuse what works and integrate AI into the tasks they already do.
Our research shows that users tend to move through distinct stages as they develop AI skills. Beginners generally try AI occasionally, often for quick answers or simple checks. Intermediate users start building repeatable habits, using AI to draft, summarise, explain, brainstorm and research. Advanced users go further — tasking AI with solving problems, analysing information, improving their work, automating recurring tasks and connecting AI to other tools.
The habits that matter most
The take-home here is that better AI use is easy to learn. Proficiency comes from a small set of behaviours that can be taught, copied and reinforced. For individuals, these behaviours include:
The biggest step-up comes when people move beyond just checking facts, and start using AI as a partner to help with generating, writing, creating, analysing and refining ideas. Beyond this, Advanced users ask AI to challenge and improve outputs over multiple rounds and are comfortable comparing between different AI tools (such as Gemini and NotebookLM) — knowing when to switch modes from fast models to longer thinking ones.
Often the easiest way for people to improve their AI use is by asking the system to help them pose better questions. This moves users from vague requests to clearer instructions, with better context and stronger outputs. Results often improve simply by explaining the task more specifically, providing relevant background information, explaining the intended audience, and setting expectations about the format or style of the response.
People should be proactive in asking their employer about AI use policies, including which tools can be accessed and what information can be shared with them. Keeping up to date with workplace guidance will help people feel more confident about using AI in their work, while avoiding risks around privacy and security.
Adoption spreads when people see examples that are relevant to their life and work. Colleagues, friends and managers can make AI feel practical by showing what they use it for, what works well, and what to avoid.
For the most confident users, creating personal agents can help them delegate workflows that carry out multi-step tasks automatically — whether monitoring information (such as weather patterns), managing routine administration (such as calendar schedules), tracking opportunities (such as job openings), or helping coordinate projects (such as organising a family holiday).
Building up the UK’s digital and AI skills base
Google Digital Garage is Google’s long-running free digital skills and training programme, designed to help people and small business owners build the practical capabilities needed to participate in an increasingly digital economy. Since launching in the UK in 2015, the programme has helped train more than 1.2 million people through online courses, webinars, mentoring and in-person workshops delivered across hundreds of locations nationwide.
Google has long been committed to connecting people across the UK to the opportunities of the digital economy. What began as a programme focused on online skills, e-commerce and digital marketing has now evolved to spearhead AI upskilling, helping to ensure that every community can harness the power of this transformative technology.
The wider significance of Digital Garage is that it lowers the barriers to digital participation. By offering free, practical and flexible training, programmes like this can help ensure the benefits of AI and digital technology spread more broadly across communities, regions and smaller businesses throughout the UK.
Workplaces are a gateway to more advanced AI use
The workplace plays a central role in exposing people to AI and encouraging them to use the technology in their everyday lives. How employers engage with AI is therefore critical. People are more confident and more frequent AI users when their organisation provides clear guidance, permission and support. Yet only a third of workers report having been given guidance that helps them use AI with confidence, and fewer than half know who to ask about responsible use.
Permission is one of the biggest practical barriers. Workers need to know AI use is legitimate: what data they can use, what tools are approved and where AI fits into their role. Where managers actively encourage AI use, adoption deepens. The share of workers using AI because their manager encourages it has risen from 23% to 32% in the past year, alongside rising demand for AI skills training.
Where clarity over permission is lacking, gender disparities widen. Half of men say they have used their own AI account to complete work-related tasks without their employer’s express permission, compared to just over a third of women. Given the evidence we have on how effective AI use improves productivity, persistent differences in adoption could risk creating new inequalities in workplace performance and career progression.
Our sector analysis reinforces that workplace conditions shape adoption. Information and communications score highest on support metrics, including management encouragement, knowing who to ask, and access to training, while retail and health & social work lag on several of these measures. We also find that trade union members are adopting AI faster than non-members, and are more positive about its role at work and beyond.
Supporting their members through industrial change
Britain’s trade unions have long helped workers adapt to economic and technological change. Working with employers and government to bring practical training into the workplace, reaching workers who might otherwise miss out.
That history of trusted, workplace-based learning is now becoming relevant to AI. Community’s partnership with Google on AI Works showed how unions can help turn new technology from something abstract into something workers feel confident using. After training, 54% of trade union participants said they would use generative AI tools more often.
Our latest research shows that British trade union members are already more likely than non-members to use AI confidently (25% versus 18%), to have received AI training at work (51% versus 32%), to have clear workplace policies and active manager support (45% versus 35%), and to believe AI will improve their career and society as a whole (40% versus 26%, and 46% versus 38% respectively).
Given their reach into sectors where adoption today is weakest — including retail, health and social care, and public administration — bringing trade unions into upskilling and AI training programmes will therefore be an important part of accelerating the rate at which the opportunities of AI spread through the workforce.
Fortunately, these gaps in outcomes can narrow quickly. Just a few small changes, such as those habits mentioned earlier in this section, can move someone from a Non-user to an Intermediate user, helping them save an average of 3.7 hours per week across their work and personal lives. That adds up to 192 hours, or eight whole days, over an entire year. It can also boost their workplace confidence and move them into a user group significantly more likely to report having been promoted in the last year.
low daily use
high daily use
Becoming a more advanced AI user also shifts attitudes. More regular and confident users become more positive about AI’s potential, both for their careers and for society as a whole — perhaps because they see direct evidence of its ability to boost productivity, advance innovation and solve other social challenges. Additionally, as workplace and personal AI use are highly correlated, targeted interventions through work will have effects that extend into the rest of people’s lives.
Business leaders can take clear steps to make a difference. Define where AI can be used, provide approved tools, train employees on regular tasks where it adds most value, and be seen using it visibly and effectively.
It should also be kept in mind that the choice facing businesses and other organisations is often not between using AI and not using AI, but between governed AI use and ungoverned AI use. Employees who recognise the ways in which AI can help them do their jobs better are unlikely simply to stop using it because policies are unclear or restrictive. Instead, many will turn to personal accounts and unapproved tools. This creates ‘shadow AI’ practices that are harder to monitor and govern, increasing the risk that sensitive information is shared outside approved systems.
When it comes to government and public services, the same logic applies. Public services can use AI to reduce administrative burden and improve service quality, but only if employees have approved tools, clear rules and confidence. Prior research by Public First found that in the UK, around two in five public servants are unsure what AI use is allowed at work, while 46% cite a lack of clear direction or guidance from leadership — factors that are stalling AI’s potential to transform public service delivery.
Accelerating AI adoption in British public services
The 2026 Public Sector AI Adoption Index shows while the UK has strong national ambition on AI, public servants are not yet consistently being given the confidence, permission and practical support needed to use it well.
The UK is already making laudable inroads in terms of public sector adoption of AI, and has pulled ahead of many comparable countries including France, Germany and Japan. However, Britain lags behind AI leaders like Singapore, and has particular gaps around embedding AI into everyday work, consistent access to AI tools across public services, clear guidance and the need to further expand role-specific training. More than half of the UK’s public servants report receiving no AI training, while many remain unclear about what AI use is permitted at work.
If the Government can provide approved tools, plain-English rules, practical training and clear routes for scaling successful use cases, it can do two things at once: improve public service productivity and help normalise safe, confident AI use across the wider economy.
Our research shows that no matter who you are, where you live, or what you do, AI can help you do more. With adoption already moving at pace, the risk with AI is not inequality of access but inequality of engagement. Upskilling Non-users and Beginners represents a lever that can help to close stubborn divides, but only if employers and government act deliberately to make it happen.
Non-users make up 10% of the population but the group is far from representative: 54% are aged 65 or over, compared with just 3% of Advanced users. Women are also overrepresented in lower-use groups, making up 59% of Non-users and 55% of Beginners, but only 41% of Advanced users. Education follows a similar pattern: 60% of Advanced users are graduates, compared with 25% of Non-users and 26% of Beginners.
The stakes are significant. The rewards of AI use rise sharply across the adoption ladder, from almost no direct benefit among Non-users to much larger time savings, higher average salaries and stronger promotion rates among Advanced users.
A workplace and sector divide is also emerging. Advanced AI use is more common where digital tools, knowledge work, and workplace permission are already established. And without deliberate action, AI risks deepening existing inequalities in productivity and career progression.
However, the divide isn’t fixed, and we have seen that even small changes can produce major benefits. Practical training, clear permissions, approved tools, and role-specific examples can move people quickly from basic to more advanced use. For example, targeted intervention by AI Works increased daily usage for women from 18% to 45% and for over-55s from 13% to 35%.
AI has huge potential to be a force for widening opportunity, but support must reach the people who stand to gain the most. By combining intuitive tools, practical training, workplace support and strong partnerships between government, employers, trade unions and civil society, the UK can help ensure more people are able to achieve their goals and unlock new opportunities for progress with AI.
Helping people ask anything on their minds
For decades, Google Search has been a gateway for people looking to find information online. Today, AI is driving the most significant transformation of Search in its history, and represents where many people will encounter AI on a regular basis.
AI Mode in Search bridges the gap between a traditional search engine and conversational AI, and people are now not just searching more, but searching differently. In addition to using Search as they always have, people can now use AI Mode to more intuitively discover new ideas and find inspiration by exploring open-ended concepts and following up conversationally. Moreover, people are using AI to ask questions across modalities — searching with voice, images, and even using Search Live to get the information they need in a natural, free-flowing conversation. More than one in six AI Mode searches are multimodal (non-text), and searches with image input are one of AI Mode’s fastest growing query types, growing by more than 40% month-over-month since launch.
Developments like these point towards a future in which Search helps people more deeply access, understand and take action on information in new ways everyday.
AI adoption is not just a domestic story. Around the world, countries are trying to address the same challenges and seize the same opportunities. And differences in outcomes are already emerging. Singapore, for instance, boasts twice as many Advanced users as the UK as a proportion of its population — offering a glimpse of what the next stage of adoption could look like.
International comparisons provide both a benchmark for the UK’s achievements to date and reasons to be optimistic about our ambition. The UK has a higher level of AI adoption than countries like Canada but has further to go to match world leaders like Singapore.
We know AI habits can shift quickly, even if that change requires outside influence. Two thirds of the British public only started using AI in the last year, showing the speed of the transformation underway. Wider global evidence also shows that national adoption can shift fast when governments, businesses and users pull in the same direction. The UK is starting in a good place, has time to catch up to global front runners — and potentially lead — if it turns broad access into better use.
A global centre for AI research and innovation in the heart of the UK
Google DeepMind is one of Britain’s most significant home-grown AI success stories, and a prominent example of how frontier AI research underpins wider economic and social value.
AlphaFold, perhaps its best-known breakthrough, has transformed biological research by predicting the structure of millions of proteins — a process that previously took scientists years of laboratory work. The breakthrough was recognised with the 2024 Nobel Prize in Chemistry and has since been used by more than 3 million researchers across over 190 countries. More than 1 million of these users are based in low- and middle-income countries, helping democratise access to cutting-edge scientific tools.
As one of the world’s leading AI research organisations, headquartered in London, Google DeepMind also reinforces the UK’s position as a global centre for AI development, acting as a magnet for top international talent, investment and scientific collaboration from around the world.
Adoption of AI is occurring at a breakneck pace. Across the UK, millions of people are getting to grips with how AI can help them work more efficiently, learn more effectively and solve problems more creatively. Over 20 million UK workers — more than two thirds of the total — have used Google tools to build their AI skills. A small share of the population have already mastered AI to a relatively advanced level, but a larger group are still in the early phases of their AI journey.
If Britain can build the habits and support structures that help people use AI well, the benefits will spread far beyond the most confident users. This report has shown the difference between basic and more advanced use is practical rather than technical: people need useful tools, confidence, permission, training and real-world examples they can adapt and implement.
For employers, the task is to provide clear guidance, approved tools, hands‑on training and visible leadership. These make the difference between occasional use and confident adoption. Where those conditions are missing, AI use does not stop — it simply moves into the shadows, fragmenting skills development and increasing risks. Supporting employees to use AI openly and properly is therefore not just about innovation, but about productivity, trust and competitiveness.
For government and public services, the same logic applies. AI can reduce administrative burdens, improve service quality and free up employee time — but only if frontline workers have the right tools and are confident in what they’re doing. That means moving beyond national ambition statements to tracking adoption where people actually work: in departments, councils, schools, hospitals and public bodies, and supporting them with shared standards, procurement frameworks and practical guidance.
Partnerships are central to making this work. Businesses bring tools and real‑world use cases. Government brings scale, legitimacy and public trust. Skills bodies, unions, local leaders and civic institutions bring reach and insight into what different groups need. Collaboration allows faster learning about what works — and quicker course‑correction when it doesn’t.
The UK’s AI opportunity isn’t just about adoption; it is also about leadership. With world‑class research capability through Google DeepMind and others, the UK can be a place where AI is not only used, but developed responsibly.
If Britain gets the next phase right — turning access into habit, habit into skill, and skill into confidence — AI will become a practical amplifier for millions of people. The challenge now is collective: to work together to make sure that opportunity is shared, sustained and built to last.
- 1Nationally representative polling of UK adults, conducted online by Public First for Google UK AI from 20 March to 2 April 2026, with a sample of 5,578 UK adults, weighted by education, gender, age, region and social grade. All results are weighted by interlocking age and gender, region and social grade to nationally representative proportions. As with all opinion polls, there is a margin of error. The margin of error is greater for smaller sample sizes, including when analysing cross-breaks of specific groups of people.
- 2Logistic regression, to understand the relationship between advanced AI use and career outcomes. Results compare advanced users to non-advanced users and control for age, gender, ethnicity, sector, education level and business size. Results are reported as odds ratios. All results are significant at P<0.05 or below.
Regression results:
- 3Segmentation of AI users, identifying different groups by frequency, confidence, depth of use and reported benefits.
- 4An AI diagnostic tool / adoption quiz, allowing people to assess their own AI habits and receive practical recommendations for how to level up.
- 5Economic modelling of Google’s product impact, including business activity, productivity, exports, consumer value and support for small businesses.
- 6Product-specific polling, covering how people use and value Google tools.
- 7Review of Google’s UK investments, tools and partnerships, including lessons from previous Google UK economic impact research, AI Works and the Public Sector AI Adoption Index 2026.