Ashwini Thota, Analytics and AI leader at Bose Corporation in an interaction with CXOToday shares his views on AI and Data Science Industry and various challenges & opportunities that are emerging across the world.
Q: Kindly brief us about the company, its specialization, and the services you provide to the organization.
Bose Corporation is one of the major makers of speakers, headphones, and other related products for the home entertainment, automotive, health, military, and pro audio markets. It manufactures a variety of consumer models for stereo systems and home theatres, including its compact wave radio system.
As a data scientist V/ principal data scientist, I develop artificial intelligence and analytical solutions to improve internal business functions at Bose Corporations. My work enhances various areas such as supply chain, marketing, sales, and more.
Q: According to you, what are some of the challenges faced by your Industry? And what is the scope of innovation?
The demand for electronics has increased significantly in recent years. Digitization of services and rampant growth of 4G/ 5G networks are some of the reasons for the organic growth of electronic devices. But one significant outlier that meteorically increased the growth of electronic devices is Covid-19. The whole world had to hit pause and prioritize remote-first culture.
The customers’ outlook towards electronics has changed fundamentally. Owning a headset was no longer nice to have; it became must have to take the work calls effectively. As a result, the global demand for integrated circuits (ICs or “chips”) and other electronic materials has increased significantly.
Because of the global supply chain shortages and the increased demand for ICs, our ability to accurately demand customer demand had never been this important. If you over manufacture a specific product, it will just be sitting on the shelves, and you wasted ICs that could have been used on other better-selling products.
If you under forecast, you lose the opportunity to sell your products. At Bose, we are obsessed with meeting and exceeding our customer’s expectations. We seek to achieve this goal by improving how we forecast customer demand to ensure we can offer our customers the right product at the right time.
While we have found traditional time series methods can help us to forecast the demand for existing products with acceptable accuracy levels, they often need to catch up in predicting the demand for new product introductions. To this end, I am actively working with my team to develop advanced models to help us with cold start supply chain demand forecasting.
Q: How are Data Science and AI evolving today in the industry as a whole? What are the most important trends that you see emerging across the globe?
Several interesting trends are emerging in the space of AI and data science. I am particularly excited about the following:
Ethical and Explainable AI
AI is often considered a black box. Deep learning models provide little or no insight into how decisions are made. Most of the research in the last decade has been on increasing accuracy and not necessarily on improving the model explain ability. As AI applications have become mainstream, organizations have started focusing on AI’s ethical and explainable aspects.
As AI products become mainstream, we will see many AI-augmented products evolve. These can be the products that you are currently using. Tasks with enormous historical data will be automated and can be offered as a feature. Propensity to buy is an excellent example of one such task. Cold calling is common in industries such as finance and insurance.
Every call a customer service representative makes is a valuable data point. All of the historic cold calls can then be used to model the prospects that are most likely to convert into an opportunity. This might fundamentally transform the job roles. In this case, the customer service agent will prioritize calling the customer with the highest propensity to buy, instead of wasting their time by making random cold calls.
Q: How are you using AI to improve customer experience and deliver relevant business outcomes?
We are using AI and advanced analytics to improve our internal business practices in marketing, supply chain, human resources, and finance, all to provide a first-class customer experience. Below is one such example:
Cold-start or new product introduction forecasting is the process of understanding the sales demand for net new products. I am working with my team on developing a robust Deep Learning approach to understanding the new product demand before launching the product. This innovative approach learns from the current product features, media spend, external market factors, etc., to generate forecasts for the new product before launch. This new capability enables us to take new proactive approaches towards manufacturing and distribution, resulting in improved customer experiences and bottom-line revenue impact.
Q: What is your Leadership Mantra?
Pace over perfection: Technologies such as AI, cloud, and data analytics are changing at a rapid pace. Every week, researchers release new algorithms, and someone is inventing new program libraries. It is vital for organizations not to have preconceived notions about certain algorithms or stick to conventional ways of doing business. The traditional development style of AI algorithms is heavily aligned with academic research, where model accuracy is highly prioritized. When it comes to building AI applications for business, leaders should be more agile and nimble and try to solve problems rapidly.
Clarity and transparency: As a technical leader, you can influence the roadmap of a product or strategy that immediately impacts your organization. You often have to coordinate with different enterprise teams to get things done. Having clarity of thought and transparent communication is very important to build trust and empower people to accomplish goals.
Q: How do you see organizations innovate with AI in the future ahead?
Data-centric AI: Increased computational power and cloud computing have enabled organizations to use state-of-the-art models to develop their AI applications. As AI products mature, organizations have realized that they need to emphasize their data more. Usually, when training a model, the data is – collected, cleansed, and transformed before feeding to the model. Data scientists then assume that the data is constant and focus heavily on modelling and building the next best algorithm. But, in reality, focusing on data quality and adding representative data can yield greater results.
We will continue to see technological innovations that can deliver a different level of personalization. Metaverse, digital twin, etc., are examples of taking personalization to the next level. AI technologies such as chatbots and voice assistants will be embedded into mainstream applications to promote the personalization of services.