India team vital to the future of flight with AI: GE Aerospace's Deshmukh

Dinakar Deshmukh, vice-president of Data Sciences and Analytics at GE Aerospace, the data science and engineering teams in Bengaluru have played a key role in building GE's digital twin concept

When OpenAI launched ChatGPT in November 2022, it marked a turning point for the artificial intelligence industry, says Dinakar Deshmukh, vice-president of Data Sciences and Analytics at GE Aerospace. The rapid rise of generative AI caught many by surprise, leaving the industry to decide whether to embrace or resist the technology. "My life went upside down when ChatGPT went public," Deshmukh said in a video interview with Peerzada Abrar. At GE Aerospace, leadership quickly noticed employees integrating ChatGPT into their work. While the company’s tech-forward culture made this unsurprising, it also raised concerns about the potential leakage of sensitive or proprietary information. This prompted GE Aerospace to create AI Wingmate, a tailored generative AI assistant designed to enhance productivity, improve customer service, and drive the future of aviation. Edited excerpts:

 

What inspired your passion for AI and data sciences?

I'm actually a core mechanical engineer. I spent close to 14 years in core engineering. I moved into the area of data science around two years back. One thing which is very exciting for me is to lead the team of data scientists at GE Aerospace.

But new data science alone may not give you the most optimised outcome. However, when you combine machine learning with domain knowledge, that's where you get a differentiated outcome. Every day there is something which is new for us to work on. That actually is very exciting. The problems are complex and the technology is moving so fast.

Machine learning and AI is not new to GE Aerospace. But my life went upside down, due to what happened in November 2022. ChatGPT became public. Post the advent of generative AI, we see that there is a significant unlock which we can do on the business side. We are applying GenAI at scale to unlock that value. But at the same time we are not losing track of conventional machine learning.

 

How has AI transformed the way GE Aerospace approaches different problems like engine health monitoring?

What changed in the last 10 years is that previously people used to see one sensor, one variable, and do a little bit of a trend plot and see if there is a deviation. What changed in the last 10 years is we started to apply machine learning where we started to look at more than five-six sensors at the same time. It is impossible for humans to even imagine what it may look like. So, we have been applying those machine learning techniques in monitoring our commercial engines.

We are able to drive significant business outcomes because of that. For example, we can improve 50 per cent plus improvement in lead time. The number of failures which we can detect through these algorithms is almost like a 40-plus per cent improvement. We reduce false positives by more than 50 per cent. All these numbers are very significant double-digit percentage improvements. That is one side of how we are applying machine learning and AI in commercial engine health monitoring. There are also areas where we are using some of this ML and AI solution to develop our digital twins, which enables creating virtual replicas of engines to monitor performance and predict maintenance needs.

 

Could you share some examples?

Let's talk about this blade inspection tool. It is almost like the way a doctor looks at your X-ray or CT-scan and says you have a potential problem. We do a very similar thing for our jet engines where we do borescope (a specialised optical instrument) inspections, and we extract images or videos from that. There are many things which are not in our control. So, what this blade inspection tool does is it will help you to get an absolutely consistent image and helps you to identify if there are any problems in that part.

Regarding the commercial engine health monitoring process, every time an aircraft engine takes off, you get information from the sensors of the engine. We apply machine learning algorithms. If there's a deviation, it'll immediately notify the human or the subject matter experts.

 

What kind of role is the India team providing for such innovations?

We have teams in India, China, London, Cincinnati and Mexico. In Bengaluru, we have engineers designing the jet engines as well as a data science team solving the problems on the same floor at the John F. Welch Technology Center (JFWTC). They partner day in and day out to drive these solutions. The data science team, in conjunction with the engineering team, is able to drive and solve significant problems. And this team played a very significant role in building this entire digital twin concept. They also put into operation and are driving the outcomes. This team has a significant role to play, especially tapping into that technology advancements which are happening in the area of GenAI. The entire ecosystem, which Bengaluru offers is also unprecedented, which you don't have anywhere else.

 

How is GE Aerospace's AI Wingmate transforming employee workflows, and how does it differ from ChatGPT or DeepSeek?

One thing which is very important when it comes to the adoption of AI by our workforce is also cultural transformation. They need to be comfortable in using these AI-based applications. When ChatGPT became public, we had our own guidelines within GE Aerospace on how to use it and how not to use it. Having said that, there was significant traffic to ChatGPT and we were concerned about a potential opportunity for IP (intellectual property) leakage. As you can imagine, we sit on very sensitive IP-related content within aerospace. To be very transparent we had to either stop ChatGPT from our GE network or provide an equivalent version of the same thing but within the firewalls of GE Aerospace where we can protect our IP, and our entire employee base can safely start using that system. So we built AI Wingmate in partnership with Microsoft to make the GenAI platform available for the entire GE Aerospace workforce. They can use it without getting concerned or worried about whether there will be data leakage. It is like the private version of ChatGPT. But what differentiates it is that this entire system is working within the firewalls of GE Aerospace. Also, we are training this system with GE Aerospace’s proprietary knowledge. We intend this to be the platform where our employees will use it for productivity, getting information at their fingertips in a fraction of time.

 

What does the future of AI-driven aviation innovation look like in the next 5–10 years, and what role will GE Aerospace play?

The recent advancements in AI are expected to provide some unprecedented opportunities to impact and influence the major parts of any typical aviation or aerospace industry. I personally think that GenAI, in conjunction with conventional AI, will transform the aerospace industry. This includes inventory planning, supply chain monitoring and product design, the way we develop our software or write code and digital applications. All these aspects are going to the next level because of this technology. We are trying to put our technology investments to work in those areas. For instance, where we have given generative AI-based coding assistance to our software developers, we are clearly seeing 15-20 per cent of productivity gains coming out of that.