AI Use Cases 

Automate and simplify data identification, aggregation, extraction, verification and capture involved in KYC (Know Your Customer) process for banks.

Predict pricing and value of real estate based on market and other information. However this is a market where occasionally large price dislocation may be driven by regulatory decisions (e.g. planning) so ML systems may struggle to capture these risks.

Automate generation of reports - for example using financial reporting to generate basic financial market analyst updates

Recognise and categorise type of damage to predict insurance claim valuation through accessing photographs provided by on-site sources. Reduces need to deploy risk adjusters - and the chance that their assessments may be influenced by external factors (e.g. interaction with customers).

Search through data and find healthcare claims that were denied by insurance companies, discovering the causes and speeding up the resolution process, ensuring healthcare systems are accurately paid.

Discover new complex interactions in the financial system to support better risk modelling and stress testing. Analysing large quanitites of data may suggest patterns that are not visible or obvious to human analysts operating with machine learning support. Equally the human will be necessary to question and improve on the ML output.

Enhance risk modelling and stress testing by discovering interactions in the financial system. This increases model robustness and strengthens it. Transparency may be an issue as the interlocked drivers become harder and harder to explain -although this has potential regulatory issues.

Predict failure and recommend proactive maintenance for mining, drilling, power generation, and moving equipment

Predict failure and recommend proactive maintenance for mining, drilling, power generation, and moving equipment

Interpolate data on ground composition to reduce the required exploratory drilling samples. This increases speed to market and reduces the potential workload.

Optimise mine extraction plans based on drilling samples, past sites, and other data including satellite and drone footage

With sufficient warning wind turbines can be positioned at optimal angles to extract maximum benefit from fast-changing gusts of wind. in concentrated fields of windmills the windflow interaction between turbines can also add to the complexity that needs modelling.

Managing event security has the benefit of operating inside a closed environment where huge volumes of data can be generated by multiple video, sensor and other data feeds. Integrating this and ensuring fast and reliable data visualisation and risk monitoring is a sophisticated AI use case.

An application of this type would categorise the food pictured in an image and send the user to an approprate recipe to enable them to create the food pictured. With the growth of social media this creates multiple opportunities.

Artificial Intelligence has become salient and understandable to the public at times when machines have beaten humans at high profile games - Deep Blue winning at chess and AlphaGo at Go. This has now extended to a series of video games. This has significantly helped the development of AI overall.

Optimise release dates and location targeting for film (or movie) launch rollouts. Typically this will involve analysing consumer data such as social media traffic and search patterns to ascertain interest and 'word of mouth' on the product.

Support multi-lingual service provision - whether to foreign visitors or minority communities - through conversational agents / chatbots. This may be deployed through a tool like Alexa in a hotel room or built in to public sector offices or online channels.

Predict product demand trends to inform production decisions. One challenge is to combine internal data sources with 3rd party (e.g. distributor) and external data sets (e.g. social media) to build a complete picture for data analysis.

Automated captioning of video (or audio) from sporting events enables the creation of sophisticated video libraries that can be searched without the need for a huge amount of (costly) human trawling to find relevant media moments. The creation of meta-tagging hugely simplifies the indexing process. The potential to then build sophisticated, targeted media feeds - for example individual player highlights - is then possible at acceptable cost.

Analyse vending machine demands and product placement opportunity for optimal product mix. This can be a key channel in specific markets and segments.

Provide clothing and beauty product proposition online based on subscriber preferences. Typically consumer fills in a questionnaire and responds to photo prompts - this is combined with other customer data, including sales, to be analysed to prompt customer product mix proposition. The customers acceptance or rejection of these reinforces the learning loop. Often there remains a human in the loop - especially initially.

Predict chance of given outcomes form sporting events based on historic data - this will include individual's performance data, team statistics and potentially broader data sets such as weather or geo-location of events. Given the array of elements that impact on sporting success the track record of success is mixed.

Inventory management / dynamic pricing to maximise capacity fill rates and return. In a (seasonally) high fixed capacity, time-sensitive business optimising this is a key driver of profitability.

An automated assistant can make a reservation for a user - initially this has been rolled out for restaurants. Natural sounding voice interfaces helps smooth the process - but does raise potential concerns about human / robot interactions. The multi-lingual usage opportunity - for example in the tourism industry - will be considerable.

Table management and reservations is a key function for hospitality providers - often accessed through multiple distribution and booking engines. Ensuring accurate and timely information is key to optimising customer throughput and satisfaction.

Optimise lane choices and path routing based on multiple data sources to reduce length, cost or environmental impact of a trip.

Optimise pricing for services or product offerings based on customer-specific data. This may include previous purchase decisions or wider data accessed from 3rd parties.

Deliver products to consumers on the move. One option is to stock a variety of products in a self-driving store - demand for the products and the highest relevant footfall location being predicted from multiple data sources. This may be the future of ice cream vans... but it is currently the combination of a lot of AI technologies so realisation may be some way off.

Driving assistant - example tasks might be auto-parking or autopilot features on highways. Both enable the driver to sit back but require continued monitoring of the situation.

Track and analyse individual players to predict likelihood of success - this has clear ramifications for scouting and talent development. Data gathered may also be used in consumer-facing applications - ranging from sports gaming to betting applications. Data gathering faces potential issues especially where it is not publicly available - e.g. training data rather than public footage - with trade-offs emerging around privacy and ownership issues.