Deloitte’s 6 most innovative AI-projects
Deloitte is working on dozens of AI-related projects the moment, both internally and for clients. We have summed up the most innovative projects in which Deloitte is using AI to make clients more effective and efficient.
Marjolein Vlaming & Ewout Bolhuis - 6 november 2017
- Alice the care robot - to meet the increasing demand for healthcare in the future, care robots are used to help elderly people live independently. Alice, a collaboration between the Vrije Universiteitin Amsterdam and Deloitte, is an example of such a robot. Last year, Alice was used for the first time in the care of elderly, and a second version is now in the making. Alice 2.0 can play a role in connecting people and devices, in translating big data and as a communications partner.
- Automatically generated Risk and Strategy reports - Since the end of last year, Deloitte has been offering its clients automatically generated Risk and Strategy reports. At DeloitteSmartReports.com, clients can order business or industry reports based on advanced and automated analyses of over 100.000 annual reports from all over the world. In less than a minute, the customer has the fully personalized report in his or her mailbox.
- TAX-I - is an AI-powered analysis tool (prototype) that provides insights into European Court of Justice case law through network visualizations, automatically generated case summaries, statistical analytics and in the nearby future even predictive models. In the hands of a tax law specialist, this tool gives them the power to form unbiased opinions on legal questions. Watch the demo below:
Product demo: Artificial Intelligence in the Tax Function
- SONAR - is a program that detects potentially incorrect (human) labelling within a database. Using advanced mathematical models, this program predicts the correct VAT treatment, applicable environmental taxes, custom code validation or other features based on the current data. If a predicted label is different from a current label, the program gives a hit.
- Fraud prediction - for this case, fraud detection is performed through machine learning. This form of AI detects fraud cases based on known fraud cases. To detect unknown patterns a sample of non-fraud cases was selected. The fraud team was presented with the outcome of the model (suspected fraud cases). In the first month 87% of the cases was fraud. Creating more examples and thus improving the detection.
- Chatbot - a change in policy or legislation can abruptly provide a large influx of requests from clients. To handle these questions efficiently, Deloitte can build a chatbot. The client only needs to provide the FAQ list with the expected questions and the corresponding correct answers. With little effort, the list is entered into the chatbot engine. After this, the chatbot can answer the incoming questions.