Best Practice AI Use Cases

While intended to assist the hearing impaired, this may have additioanl widespread applications. The technology must work in real-time and may be used on portable devices, which adds the challenge of limited available compute power.

Image analysis or GCMS analysis in a high throughput manner. Applications of GCMS include drug detection, fire investigation, environmental analysis, explosives investigation, and identification of unknown samples.

Automatically generate content for marketing materials

Robo-advisor automates investment and portfolio rebalancing and transactions based on defined client needs, responding to real-time market data

Automate personal finance management (e.g. loan repayments by personal agents). This might be a consumer-downloadable app that syncs with their banking products to support achievement of personal financial goals. This might include automating savings for consumer by forecasting wallet availability based on customer spending habits and preferences. A typical use case deployment is a "rounding up" function on a digital payment - so a $3.63 payment triggers a round up to $4.00, automatically depositing 37 cents in a specified savings account.

Optimise blend mixture and process timing of raw materials being used in refining and similar processes. Sustained small marginal improvements can have a significant cost impact - especially where human involvement can be costly or potentially hazardous.

Emails that are tailored to individual behaviour

Compile aggregate analyst report sentiment score based on the balance of positive to negative words in written reports

Predict and promote the next appropriate offer for a customer to maximise chance of driving desired behaviour (e.g. sales, click-through, retention offer acceptance etc).

Analyse candidate patient data to measure probability of individual resistance to deployed drugs, typically projected over time and across a broader population.

Assist developers by using AI to help them accelerate code development by automatically spotting errors

Censor user generated content on social media and other platforms. or This may be used to reduce hate speech or incitements to violence similar negative issues - but can also be used for political censorship. The line between the two can become a matter of debate.

Monitor and analyse interactions with customers (potentially across all channels ranging from market research to social media to direct contact) to create insights to improve product offering. The aim is to ensure that there is a positive feedback loop in to product development.

Optimise medical product (devices, robots, drugs etc) launch strategy based on past launches and relevant data

Prediction modelling typically used to support key decision-making steps in the judicial system - e.g. sentencing and early release. Historically rules and expert based systems there remains perceived risk of bias (e.g. on ethnic grounds) and lack of transparency and accountability.

Measuring key physical indicators - for example motion and gate - to diagnose potential medical complications

Sales reps actions and leads will be analysed according to sales process and potential customer response to suggest the next best action

Sending out street cleaning teams in response to grafitti or rubbish that has been identified using street and mobile cameras rather than simply deploying in regular sweeps. This ensures that teams respond more swiftly to emerging problems. A similar approach can be applied to tending street vegetation, especially trees.

Continual research and observation of the drug's effects on patients after the drug becomes generally available

Monitor application KPIs and performance for resource management to identify issues before they become widespread or too costly

HIgh quality candidates may not be chosen for the roles that they applied for initially. CVs that are kept "on file" can be triaged for different roles.

In environments of high complexity and potential risk being able to label components automatically can help with mutliple operational tasks.

Construction work is very reliant on the quality of both staff and site management, roles which may be harder to deliver in hard to access or inhospitable locations. Machine vision can be used to ensure that quality standards are being met and errors minimised and to ensure a rapid feedback loop to avoid potential cost (and engineering safety) issues.

Evaluate customer credit risk using application and other relevant data to improve accuracy and reduce unconscious bias in real-time underwriting decisions. Signals may vary - for example signs of hesitation whilst form filling may indicate a higher propensity for fraud.

Predict risk of churn for individual customers or clients and recommend intervention strategy - this may be involuntary (ie due to bankruptcy) or voluntary (ie switching accounts) churn

Monitoring of body (eg iris photos or skin conditions) to predict potential for issues of medical concern. May be time relapse dependent (eg size of mole over several days) or point-in-time diagnosis (e..g iris photos). Can be deployed as an app in some circumstances.

Provide support for medical teams to triage incoming patients relative to their medical symptoms and history and personal situation. Data used may include audio, video, comparable, historic and real-time data.

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.

Robotics can automate many route activities that typically require human actions. This is inevitably a very broad use case and the requirements will vary - from reducing potential physical risk to humans, to maximise process uptime or simply to reduce variable costs. These will drive different configurations of project.

Determine where brand is being used appropriately and inappropriately.