TechTalk Audio: Responsible AI in automotive

How do we make self-driving cars safe? And what do we need to look out for if we want to use AI in public transport? Rudraksh Bhawalkar, Global Delivery Lead for Accenture's internal Responsible AI program, addresses both these questions and more in the fifth episode of TechTalk Audio on Responsible AI.

Tobias Imbach spoke with Rudraksh Bhawalkar

To the point:

- In the context of autonomously driven cars as well as public transportation, there are many use cases which show how important responsible AI is.

- In both areas, if a machine makes a mistake, it may cost people their lives. Thus, they have to be explainable and transparent – no black box behavior.

- For example, imagine an autonomous car is trained on the wrong data: it learns with traffic data from a rural area and is later used in a big city – the car might drive too aggressively, because the AI thinks that based on the data, there is no traffic.

- Despite training a vehicle with data, some situations in traffic are unpredictable, e.g., a person just walking into the street without a zebra crossing.

- If a person suddenly crossed a street, a car would need to identify the object on the street, calculate the outcome and take the right decision to save both the person in the car and the person outside. This depends very much on object identification and image recognition.

HOW DOES AI MAKE ROADS SAFER?

For cars and public transport vehicles to be able to drive safely at all, they need well-maintained traffic routes. But road management is complex, and there are usually not enough skilled workers. How can modern technology help here?

The start-up Vialytics has developed a solution that uses artificial intelligence and computer vision to detect and classify road damage. How this relieves the public sector, and to what extent this solution can also be used in disaster situations?

Find out in this podcast!