Applications of Artificial Intelligence in Business

In May, Google demonstrated the ability of its artificial intelligent (AI) agent Duplex to have an actual conversation with real life people. It demonstrated it could book a hair appointment but struggled with a more nuanced conversation when attempting to make a restaurant reservation.
 
Whilst there is a lot of hype around AI and a lot of work to be done before an agent passes the Turing Test, the impact AI is having on business should not be underestimated. Voice controlled digital assistants and facial recognition in smart phones are just the beginning.
 
Research firm Tractica estimates that global AI enterprise software revenue will grow from $644 million in 2016 to nearly $39 billion by 2025. For enterprises integrating AI learning techniques into existing business process platforms and case management systems the value moves beyond cost savings in back office automation towards better better marketing and customer experiences.
 
However, the practical application of deep learning methods requires thousands of data records. It stands to reason that AI will have most impact on organisations and business functions with access to large data sets in the near term. Here are 3 examples of where AI is making an impact on business today.
 
Practical applications of AI in business
 
1.  Analytics in the marketing department

Marketing personnel deliver everything from creative content creation to analysis and reporting, much of which is time-consuming, repetitive and painstaking administrative. AI is already having an impact on marketing departments that use AdWords. Even the smallest organisation can take advantage of Google’s vast data sets to optimise bids.
 
In the past, marketers using AdWords would traditionally have spent a lot of time analysing multiple advertising factors; time of day, ad copy, keyword targeting, device, location to help them optimise campaigns. Now AdWords uses machine learning to do this administrative job for them. It analyzes countless signals in real time to reach consumers with more useful ads at the right moments.  Google’s machine learning technology optimises bids for visitors who are more or less likely to convert.

The next step change for the marketing department will be using the AI to improve creative copy content. Combining the inbound marketing lead data with sales data to determine the best campaigns for high quality leads will improve quality and increase win rates. 
 
2.  Detection in the fraud department
 
According to tax and advisory firm Crowe Clark Whitehill and Portsmouth University fraud costs the UK £110 billion and £3.2 trillion globally. Traditional in house fraud detection teams do not have the resources to monitor and mitigate the growing volume of fraud attempts on a global scale.
 
Credit card companies have vast datasets including chargebacks and fraudulent purchases. By finding connections and patterns across types of purchases, locations of purchases, and types of customers, departments  can use the “labelled” instances of fraud to predict other transactions that are most likely to be fraud. Machine learning applied to vast amounts of data can help transform workflows and outcomes so that businesses can stay ahead of technologically advanced criminals.  
 
Organisations like Mastercard are already using machine learning globally to deliver enhanced fraud score on every transaction on their network. The technology improves the accuracy of real-time approvals and reduces false declines.  What’s revolutionary about the ever evolving fraud detection is that it is happening in real time, so that fraud can be caught instream rather than afterwards, reducing losses and saving costs.
 
3.  Answering questions in customer services

Gartner predicts that by 2020, more than 85% of all customer support communications will be conducted without engaging any customer service representatives.
 
Machine learning is being applied to answering rote customer questions. Using historical customer service data, natural language processing and algorithms that continuously learn machine learning can improve customer service by providing speedy and accurate responses.
 
A good example of this might be ‘suggested customer responses’. Automated responses based on social messages are being used to answer customers questions more quickly. For example, when you message your airline on Facebook to tell them you’ve left your phone on the plane, it’s likely the response has been suggested via automation to provide a more consistent and accurate response.
 
The AI can identify the query is:  i) question, ii) neutral in sentiment, iii) urgent, iv) post flight and v) a case relating to a lost item. This suggested response leaves human customer service agents more time to deal with the difficult questions and handle exceptions. It also means customers can have their queries answered 24/7 because bots never sleep.