AI Use Cases 

It is a matter of debate what the nature of creativity is. If it is to gather multiple inputs and then to combine these to deliver a new take - in this case as a piece of art, whether painted, digitally created or even scuplted with basic robotic functionality - then this is creativity. If creativity requires emotion or understanding then this is not it.

Automate generation of conversational media content for audio driven interactive programming on conversational platforms (e.g. Alexa). This is for entertainment purposes.

Vehicle cybersecurity is a hugely growing area of focus both because of large-scale risks but also the threat of targeted attacks. Potential spin-off criminality include blackmail, terrorist activity and activity by hostile state actors.

Identify vermin and propose appropriate treatment approach - for example this can be deployed as an app using Google's image classification

Optimise in-store product assortment to maximise sales. Store assortment choices may vary by location, customer demographics and seasonality for example.

Identify pertinent features like copyright infringement or audience age suitability in media content. This will support deployment of control tools (eg deletion) especially on high volume video viewing sites.

Recognise complex voice commands to access a greater variety of services and features. Issues might include multi-demands in a single sentence as well as ambient noise - voice recognition then translates to NLP to system commands to ensure job is done.

Self-driving vehicles are a poster child for AI. In fact there are multiple levels of risk, autonomy and functionality that this encompasses. There will likely be many linked network vehciles or protected environment deployments before we see fully autonomous vehicles deployed in a non-test environment on city streets.

Personalise in-vehicle suggestions based on location data and passenger requests. This might include recommending local attractions, garages for proactive maintenance and finding parking places for drivers.

Using AI to automatically classify businesses for inclusion in a corporate listing database. Key financial and business data will be categorised at the same time. This may be an initial step in building an investment process where other steps may also be automated (potentially with AI).

Software which automates the work of human moderators to identify abuse and misconduct in online messages between video game players on community boards.

Generate research for news content - stories may emphasise AI (negatively or positively) or simply be data journalism on large data sets

Being unable to find a parking place can be a nightmre for drivers - and a missed revenue opportunity for parking space providers (and an issue for the communities that they operate in). Using AI to synthesise multiple data sources and predict space availability is a strong use case for optimising what was not historcially considered a market for practical purposes.

Potential optimal store locations can be modelled based on data fields such as footfall, customer demographics and consumption patterns, competitor siting and so on.

Companies use AI to automatically flag and remove user content on social media and other platforms which violates regulations and government mandates

Predicting risk of injury from personalised data monitoring and big data analysis of historic activity can help athletes and their teams minimise activity liable to cause problems and to initiate mitigation strategies. Potential legal issues around privacy, data ownership and liability need management but this could have a positive impact on teams and individuals.

Predict failure and recommend proactive maintenance on vehicle components. This will be increasingly important for services such as automated transport-on-demand business models.

Enhanced customer's retail experience and service through chatbots able to provide access to wide ranges of knowledge at scale.

Leveraging machine vision to enable customers to search products by image or video to immediately access the information they need.

Discover new trends in consumption patterns (e.g., viral content). Increasingly this sort of data-driven approach is driving editorial decisions - potentially to the detriment of risk-taking originality.

Promos/upgrades/offers are a key part of the travel industry's customer relationship (and capacity utilisation) approach. Optimising these with ML reduces risk of mis-allocated (or mis-aligned) incentives.

Personalise advertising and recommendations to target individual consumers. This might be based on a wide spectrum of data (mobile, social media, location, etc.) but much will be based on website analytics and tracking widget deployment.

Playing poker has an additional level of information challenge to other games that AI has conquered - namely the element of bluff.

Identify and navigate roads and obstructions in real-time for autonomous driving. This requires a mapping of both relatively static elements - such as the road layout - but also dynamic threat assessment.

Workforce management (also applies to lots of verticals) - hospitality industry workforce is frequently characterised by high turnover, casual working and seasonal demand. This complicates the challenge of maintaining performance.

Sports betting is a hugely growing area, where increased data volumes on individual, team performance and market statistics means machine learning can build betting odds for bookmakers - and also for bettors to attempt to game the odds offered. Some areas - reading human emotions mid-game for example - remain harder to access reliably. This creates the opportunity to offer more targeted products (for example: chances of a given player scoring in the next x minutes).

Autogenerate captioning on video or audio files. This speech to text service potentially saves significant human effort but will work less well in languages that have not yet had the full research effort of - say - English and Chinese, where error rates remain a challenge.

Using personalised taste and health feedback machine learning can help recommend (and, in the future, produce) personalised dishes. More broadly it can help recommend a personal diet plan with further inputs, e.g. seasonal or regional foodstuff availability. Tying this to shopping, food distribution or travel recommendation channels helps close the supply loop.

Predict viewership for new content to optimise editorial investment decisions. The focus may be on single items of content or wider strategic approaches to content generation.

Use video, images or robots in the retail area to audit and analyse shelf space usage. Optimising shelf space drives retail sales, can reduce risk of running out of stock and it can also be use to prioritise shrinkage risk areas.