More than the headlines about chatbots would suggest. AI-assisted analysis of X-rays and MRIs helps clinicians spot anomalies – a lump or irregularity the human eye might miss – and the same pattern-recognition is applied to EEG, MEG and ECG scans to help detect potential issues in the brain, muscles or heart earlier than manual review alone.
On the research side, AI can process thousands of medical papers in seconds, which is less glamorous than diagnosis but arguably more valuable at scale. Consumer-grade wearables now use the same underlying techniques, analysing heart rate, sleep patterns and blood pressure continuously rather than at a single annual check-up, and personalised-medicine applications are starting to estimate the likelihood of developing certain conditions or how a patient might respond to a specific drug.
By PwC's widely-cited estimate, AI could contribute up to $15.7 trillion to the global economy by 2030. That is a large, round, frequently-repeated number, and it is worth treating as an order of magnitude rather than a forecast anyone should bank on.
The honest caveat: PwC produced it, and every consultancy – including this one – has an incentive to make the number sound big. It remains the most-cited figure in the field, which is itself informative about how the opportunity is being priced by the people actually allocating capital to it.
Both, and anyone offering you only one half of that answer is selling something. On the creation side, AI is projected to create 133 million new jobs globally by 2030.
The jobs that disappear and the jobs that appear are rarely the same jobs, held by the same people, in the same places.
Alchemy ConsultingThat is a job-creation figure, not a net one, and it sits alongside a well-documented parallel story of displacement in roles built around repetitive, rules-based tasks. Which is precisely why the transition matters more than the headline total.
Closer than it feels, on current projections: an estimated 33 million autonomous vehicles are forecast to be on the road by 2040, built around sensor arrays positioned front, side and rear, and computer-vision algorithms doing the interpretation.
You are almost certainly using a lighter-touch version of the same technology already. Google Maps and Waze combine historical traffic data, real-time updates from other users and machine learning to predict your journey time – which is AI quietly doing useful work well before anyone hands over the steering wheel.
Almost everywhere, and almost invisibly.
Facebook and Instagram decide what appears in your feed based on interactions, interests and even how long you pause on a piece of content. None of this required a press release; it simply became the plumbing.
Largely through pattern recognition operating at a scale no manual process could match. Fraud-detection systems use outlier detection techniques to flag irregular credit card transactions, and banking systems catch unusual patterns – a purchase suddenly appearing in another country, say – far faster than manual checks ever could.
The scale requirement is the real story: anti-fraud systems need to operate across every banking transaction, every credit card transaction, any moment, anywhere in the world – not a job-spec any human team could fulfil unassisted.
Natural language processing, mostly. AI-driven customer service chatbots use NLP to parse a query and return instant feedback rather than a hold queue. It is not the most exciting application in this list, but it is one of the most widely deployed – and the underlying technique, understanding intent from ordinary written language, is the same one doing more sophisticated work elsewhere on this page.
Often in unglamorous asset-management roles that never make the news. One Australian electricity provider used AI to identify and verify potential faults in electrical lines through image processing, saving engineers countless hours previously spent manually reviewing thousands of aerial images.
That is the pattern worth noticing: the most defensible AI business cases are rarely the flashiest ones. They are the ones that quietly remove a genuinely tedious task from a skilled person's week.
Yes, through precision agriculture. AI systems analyse multispectral images to detect nutrient deficiencies or pest problems before they become visible to the human eye, and combine weather forecasting with guidance on when to water crops and which pesticides to use.
It is one of the less-discussed AI applications, largely because it happens in fields rather than boardrooms – but it is a genuine example of AI improving a resource-constrained, high-stakes decision that has historically relied on experience and guesswork.
Figures drawn from PwC's global economic impact study, and reporting via Medium/Design Bootcamp, University of Cincinnati Online, and University of Hull Online, as of September 2026. This is a living register, not a one-off report – it will be revisited and refreshed each year as the underlying numbers move. Companion piece to "The AI Capex Wave in Numbers."