In an exclusive interview with Ankur Pawa, Senior Vice President – Digital, Sasken technologies, the veteran pins the calibrated trends of AI and ML. Pawa underlined, AI and ML are both a potential friend and victim of Cybersecurity. To further undercut on the market, Pawa also shared Sasken Technologies’ focus and solutions and their role to propel the AI and ML future. Edited Nub.
- Machine intelligence is progressing at an astounding rate. The technologies like AI and ML are taking over the industry worldwide. What according to you are the positive and negative outcomes of it?
AI provides organizations with an unprecedented opportunity to drive growth, reduce cost and improve customer experience. According to a McKinsey survey, this translates into an economic value potential of $3.5-5.8 Trillion – such is the profound potential impact. The application of AI range from autonomous vehicles, healthcare diagnostics, supply chain optimization, predictive maintenance of machines, customer acquisition and churn management, etc. As humans and consumers, this translates into greater convenience, higher predictability, improved choices and superior health care in our daily lives.
While these benefits exist and significantly outweigh the negatives, AI is a double-edged sword which comes with its potential risks and challenges. Cybersecurity threats continue to be a big concern for AI. Laws and regulations governing the usage and applications of AI need to evolve to address the issue of culpability in case of an AI-linked mishap. Usage of AI for malicious intent whether it is fake news or cyber fraud is also a big risk that needs to be addressed.
- What trends do you forecast for AI and ML in 2019?
Here are some of the trends that we see in the industries we operate in:
- AI/ML adoption in the Industrial and manufacturing world will see a massive surge. 2018 was a year of creating the foundational IoT systems to capture the sensor data. 2019 will see this data being used to drive several AI/ML use cases around asset health and performance management, shop floor planning and improving the OEE (overall equipment effectiveness).
- Explainable AI (black box to glass box algorithms) will garner more interest, investments and mainstream adoption. This stems from the fact that models like deep-learning behave like black boxes and fail to provide an explanation for predictions made by the model. For example, if one was using ML to determine whether a loan application should be approved or not and the model recommends a rejection, then we cannot accurately determine ‘why’ the model has rejected it. Some regulated industries and other specific use cases require this causal analysis and hence we are seeing a lot of attention towards this.
- Personal assistants will make their way in the B2B world. The field force and the shop floor operators will get AI/ML driven personal assistants that will help them do their jobs better. The agents will provide seamless access to relevant information over voice commands and also help in resolving complex issues and troubleshooting.
- Adding to the recent trends, AI and ML are said to change the scenario of every sector. Which sectors are the early adopters of these nascent technologies and what impact shall it bring given the understanding of current business processes?
Retail/CPG, Banking and Financial Services, Telecom and Logistics are the largest users and adopters of AI/ML. One of the biggest applications of AI/ML has been in consumer sales and marketing and hence the B2C industries have led the adoption curve there. We also see great use of complex AI/ML in the Automotive sector for autonomous vehicles. The next wave of adoption is going to be driven by the Industrial and manufacturing sector as they go on to collect high volumes of quality data.
- The vast use of data also means that the security would be an issue. What are your thoughts about the impact of AI and ML on Cybersecurity?
AI and ML are both a potential friend and victim of Cybersecurity. AI/ML algorithms are used to predict and detect cyber-attacks and malicious activities making it a great aid for cyber security. On the other hand, there is great concern on AI algorithms being hacked. This hacking could hurt in multiple ways – it can potentially change the behavior of an AI/ML model by forcing biases thereby driving malicious actions. It can also reverse-engineer the model or the training data by analyzing the patterns of responses it generates by generating large numbers of interactions with the algorithm. This can pose great threats to privacy and amount to stealing of IP.
- Lastly, how closely is Sasken gripping on these technologies?
We are investing across the data value chain to ensure we provide complete solutions around AI/ML to our customers. In terms of specific areas and solutions, we have built our own reference Big Data architectures for different industries that have varying needs and requirements. We have also invested heavily in the area of AI and ML and have built several solutions. For example, we have created an Advanced Image Recognition solution which helps in drowsiness detection or distraction of a driver, fire detection in public places, detecting pedestrians on the road, etc. Another interesting example is a solution we have built in the areas of Asset Performance Management and Digital twins which helps us predict the health of machines and prescribes maintenance activity so that the remaining useful life of the machines can be increased while mitigating any risk of accidents.