← Back to the homepage
BIGGER PICTURE 99 AI stories

NASA parts shaped like bones.
Burnt scrolls read unopened.

Explore all 99 applications
Three metal telescope supports with branching, curved struts and bolt holes on a workbench.
NASA / Henry DennisNASA media guidelinesWhy this shape?
How this collection was selected

We selected 99 distinct applications with a traceable benefit for people or a demonstrated step towards one. Each links to an original study, a report by an involved institution, or an explicitly labelled provider report. The order starts with particularly tangible examples; it is not a ranking.

“In use” describes use in the documented project, not independent proof of effectiveness or a guarantee of current operation. Study evaluations and research are labelled separately. Better predictions, higher detection rates and saved working time do not automatically mean lives saved or a net social benefit. Each case explains its specific limit. This selection is not a statistical comparison with incidents in the register.

Sources reviewed: 14 November 2026

How did they do it?

Every example freely accessible

13–24 of 39 matches · 99 applications in total

AI-generated illustration: Layered scans and organ outlines illustrate the contouring work before radiotherapy.
Before aiming at a tumour, map the healthy organs that radiation should spare.
Better careIn use

The invisible drawing work before cancer treatment

Discover the story
Result & evidence

OSAIRIS outlined healthy organs on treatment scans before the oncologist reviewed them, reducing repetitive preparation work and helping cancer teams move patients towards radiotherapy more efficiently.

Documented result

The 2024 evaluation found 36% less contouring time for the prostate model and 67% less for head-and-neck organs, including clinician editing.

Cambridge University Hospitals; Microsoft Research
Cambridge, United Kingdom 2024

Oncologists checked every segmentation; faster planning alone does not quantify improved cure rates or actual waiting-time reductions.

Reviewed 2026-11-14
AI-generated illustration: A conceptual cheque verification process separating suspicious duplicates.
A cheque leaves two records. A suspicious difference can become an investigator’s clue.
Public & everyday servicesIn use

Treasury says AI helped recover $1 billion lost to cheque fraud

Discover the story
Result & evidence

The US Treasury used machine learning to identify suspicious government cheques faster. Earlier investigation and recovery help protect public funds that support legitimate recipients and public services, rather than leaving losses undetected.

Documented result

Treasury attributed $1 billion in fiscal-2024 recoveries to faster cheque-fraud identification using machine learning.

US Department of the Treasury
United States 2024

Agency-reported recoveries; the broader $4 billion programme total cannot be attributed entirely to AI.

Reviewed 2026-11-14
AI-generated illustration: Conceptual wind turbines connected to abstract future power-supply blocks.
The next day’s wind becomes a promise the power grid can plan around.
Energy & climateIn use

Making tomorrow’s wind power easier to promise

Discover the story
Result & evidence

A neural network forecasts wind-farm output 36 hours ahead and recommends delivery commitments, helping renewable electricity fit power-market schedules and making variable generation more useful to grid operators and its owners.

Documented result

On 700 MW of US wind capacity, Google reported roughly 20% more economic value than a baseline without time-based grid commitments.

Google; Google DeepMind
Central United States 2019

Higher economic value does not mean 20% more electricity or a measured 20% reduction in emissions.

Reviewed 2026-11-14
A world map from NASA’s LHASA model marks areas of landslide exposure using red, orange and yellow.Source illustration
NASA Scientific Visualization Studio / Zoey N. Armstrong and Helen-Nicole Kostis; science: Thomas A. StanleyNASA media-use guidelines
Disaster preparednessIn use

Which rain-soaked slope deserves attention first?

Discover the story
Result & evidence

NASA’s machine-learning model combines rainfall and ground conditions to flag increased landslide likelihood, giving emergency managers a consistent overview when scattered mountain hazards are difficult to assess on the ground.

Documented result

LHASA supplies kilometre-scale hazard maps every 12 hours and was integrated into the operational DisasterAWARE platform in 2023.

NASA Goddard Space Flight Center; Pacific Disaster Center
Global 2023

Regional hazard probabilities guide attention; they do not establish whether an individual hillside is safe.

Reviewed 2026-11-14
AI-generated illustration: Conceptual small marsupial near a camera trap in regenerating burned forest.
Who returned after the fire? Millions of camera-trap images hold the clues.
Nature & foodIn use

Finding a tiny survivor in millions of forest photographs

Discover the story
Result & evidence

AI sorts camera-trap photographs after Australia’s bushfires, helping land managers spend less time labelling images and use wildlife evidence to plan habitat recovery, invasive-species control and future fire management.

Documented result

Eyes on Recovery partners reported image processing five to ten times faster, with one 500-hour workload reduced to about 100 hours.

WWF; Conservation International; Google; Eyes on Recovery partners
Southern and eastern Australia 2022

Faster monitoring informs restoration; it does not prove AI caused wildlife populations to recover.

Reviewed 2026-11-14
AI-generated illustration: Alternative train movements compared in a conceptual railway simulation.
One late train, several possible futures. Dispatchers can compare the consequences.
Public & everyday servicesIn use

Try the next train decision before making it

Discover the story
Result & evidence

Deutsche Bahn’s AI-Dispo simulates the live rail situation and suggests actions to dispatchers. It helps staff spot conflicts early and choose responses that limit delays for passengers on busy suburban networks.

Documented result

Deutsche Bahn reports that AI-Dispo avoided around 95,000 train-delay minutes during 2025 across its pilot operations.

Deutsche Bahn
Stuttgart, Rhine-Main, Munich and Berlin, Germany 2025

An operator-reported pilot result; saved train-delay minutes are not the same as passenger-hours saved.

Reviewed 2026-11-14
AI-generated illustration: A shorter computational path organising a sequence of blocks.
Even sorting three numbers still had a shortcut waiting to be found.
Scientific discoveryIn use

One fewer instruction, repeated across countless sorts

Discover the story
Result & evidence

AlphaDev discovered more efficient sorting routines that were incorporated into an open-source C++ library. Developers can benefit from improvements to an everyday computation without manually reinventing the low-level algorithm themselves.

Documented result

Reported sorting gains reached 70% for short sequences and around 1.7% for sequences longer than 250,000 elements.

Google DeepMind; LLVM libc++ community
Global 2023

Specific routines and benchmark conditions; these percentages are not speedups for entire applications.

Reviewed 2026-11-14
AI-generated illustration: A photographed everyday scene becomes an interactive description.
The useful detail may be the one a fixed image description leaves out.
Accessibility & independenceIn use

A photograph becomes a conversation you can steer

Discover the story
Result & evidence

Be My AI described photographs and answered follow-up questions for blind and low-vision users, providing an additional way to obtain everyday visual information through a phone.

Documented result

Interviews with 14 visually impaired users documented practical image exploration and valued follow-up questions, alongside errors and a need for human checking.

Be My Eyes
Global 2025

The source documents deployment and user testing, not a controlled independence outcome; human volunteers remain available to check descriptions.

Reviewed 2026-11-14
AI-generated illustration: Public-response cards organised into an inspectable map of themes.
Two hours to sort the responses. Another 22 for experts to check them.
Public & everyday servicesIn use

50,000 public responses became a map analysts could inspect

Discover the story
Result & evidence

Consult groups large volumes of consultation responses into themes for public officials. It reduces repetitive sorting work so specialists can spend more time interpreting what citizens and organisations actually submitted.

Documented result

For the Independent Water Commission, over 50,000 responses were grouped in about two hours, followed by 22 hours of expert checks.

UK Department for Science, Innovation and Technology
United Kingdom 2025

Human checking remains necessary; quick classification does not prove that every minority viewpoint was represented fairly.

Reviewed 2026-11-14
AI-generated illustration: Conceptual converted factory with sensors controlling room heating before dawn.
An old heating system can learn new timing without a new boiler.
Energy & climateIn use

Teaching an old factory when to start warming up

Discover the story
Result & evidence

AI adjusts ventilation and water-system operating schedules in an occupied Milan business park, helping building teams avoid unnecessary equipment use while responding continuously to changing conditions and comfort needs.

Documented result

The project team reported 183,300 kWh of electricity savings, or 26.1%, on controlled equipment from February 2022 to August 2024.

BrainBox AI; R2M Solution; RealStep
La Forgiatura, Milan, Italy 2024

Savings cover the controlled equipment, not the entire campus; the report was written by project and provider staff.

Reviewed 2026-11-14
AI-generated illustration: Conversation is transformed into readable lines on a phone.
Spoken words become something a deaf customer can read, then answer.
Accessibility & independenceIn use

A shop conversation becomes readable on a phone

Discover the story
Result & evidence

Live Transcribe used speech recognition to display conversations on a phone, helping deaf and hard-of-hearing people follow interactions in shops, tours and other everyday situations.

Documented result

Co-developer Gallaudet reported support for over 80 languages and described daily use to understand shop staff and reply in text.

Google; Gallaudet University
Global 2023

Availability and individual experiences are documented; this source does not establish transcription accuracy in every language or setting.

Reviewed 2026-11-14
AI-generated illustration: A suspicious payment paused for review before funds leave.
The sender is real. The danger may be waiting at the other end.
Public & everyday servicesIn use

The bank sees a scam before the money leaves

Discover the story
Result & evidence

TSB uses AI risk signals to identify suspicious bank transfers, including payments people are manipulated into authorising. Bank staff can intervene before money reaches a scammer, adding another layer of protection for customers.

Documented result

Mastercard reports that TSB’s detection of authorised push-payment fraud increased by 20% after adopting Consumer Fraud Risk in July 2023.

TSB; Mastercard
United Kingdom 2023

Provider-reported detection improvement, not a randomized estimate of losses prevented; false alarms still require review.

Reviewed 2026-11-14
A bigger picture of AI starts with concrete evidence.Back to filters ↑