Listen to this story
Despite the never-ending benefits of AI and analytics, many companies still struggle to adopt them to scale their business, know customers better, and add value. According to experts, the adoption has been slower than expected.
At the recent Scale TransformX Conference, Google Cloud chief Thomas Kurian, alongside Scale AI founder Alexandr Wang, shared his insights into the state of AI adoption, vision for the company, business philosophy, latest innovations, multiple use cases and more.
Addressing the barriers to AI adoption from Google Cloud perspective – which has been one of the early movers in AI – Kurian said one of the barriers is rarely about the algorithm itself as it is very different across industries.
Subscribe to our Newsletter
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Citing the retail industry, particularly in line with the recommendation system used to make product discovery, Kurian said that the biggest challenge is standardising the meaning of the product and the catalogue because unless you have a standardised definition of the products and the data behind the algorithm is clean, it is super hard to get a recommendation.
“We worked with H&M, Macy’s, IKEA, and Bloomingdale and for a large number of these brands, a big part of the problem is how you label and clean the data upfront and standardise it before you get into the algorithmic phase,” he added, saying that is the one part of things they see.
“The second part is that for large organisations to adopt AI, they have to need to input the results of the algorithm back into their core processes,” said Kuran, citing GE (Europe), which they are helping manage the grid and produce electricity effectively using AI and analytics solutions.
“The third is the people’s side. There is change management you go through to get people to trust the algorithm,” said Kurian, giving examples of banks that can disburse loans, mortgages and offer financial products and services. First, however, it requires customers to get comfortable with it, where things like fairness and others come into play.
“Often, when people look at AI, they think it is a skills issue. Sure, there is not enough talent in the ecosystem. But, things are getting easier as the models get more and more sophisticated,” said Kurian, adding that people often forget about these other important issues in adopting.
“In the past three years, we have seen a huge ramp up in our business, and the credit goes to all the people who joined us,” said Kurian. Roughly half of Google Cloud’s employees have been hired since 2020. Still, they all did an amazing job together, he quipped.
Throwing light on the application of AI in different domains, Kurian said they recently worked with a large financial institution in Hong Kong and Shanghai Bank, which used its machine learning solutions to detect fraud. “There are a lot of false positives, where people are doing something called anti-money laundering, which our AI algorithms can be super-precise in detection,” he added. Further, he said, since regulators are involved in the process, explainability becomes a big deal.
Citing Renault, he said that they help the company monitor all of their factories, where they process roughly a billion datasets every day. “Obviously, humans can not process that,” said Kurian. Google Cloud has also helped IKEA build a recommendation system. Kurian said that people shopping for furniture and products are not the same in many countries as they have different buying habits. “Those are all the things we have learned applying our AI in different contexts in different parts of the world,” he added.
Ever since Kurian took over as the CEO of Google Cloud in 2019, the company has witnessed meteoric growth that has tripled over the past few years. As per its fourth quarterly results, the company has an annualised revenue run rate of $22.16 billion, up from $19.96 billion based on its third-quarter results.
Sharing the reason for its incredible growth, Kurian said everything in the world is becoming a software-powered technology industry. Citing the automobile industry, he said that vehicles are becoming more software than mechanical.
Similarly, telecommunication companies rely on platforms to deliver applications and manage their network effectively. Banks, on the other hand, are becoming more and more virtual each day, and all of the products are based on data, and how they use them to their advantage and enhance customer experience. This is redefining the banking landscape. “Ask yourself, when was the last time you visited a branch of a bank,” quizzed Kurian, saying that a lot of work at Google Cloud has been about pushing the technology innovation far, and making it super easy for people in different industries to adopt, access, and scale.
“We offer every part of the stack that we have from the hardware network to software abstractions, to things that are more packaged because different organisations have different levels at which they have the expertise and want to adopt technology,” said Kurian.
Google Cloud’s AI & analytics strategy
“Our vision is super simple,” said Kurian. Speaking of smartphones and how it was able to bring various functions of computing, camera, communication and the internet into everyone’s pocket, he drew parallels, and noted how Google too is trying to take all the technological innovation and make it super simple for everyone to consume it.
This explains its investment in global data centres and the development of new types of hardware and large-scale systems, alongside working on software to help customers handle high-scale computations, tools for data processing, cybersecurity, and machine learning, etc.
Kurian said machine learning and AI has done much work in the past three to four years. “We look at our work as four elements,” he added. The first and foremost include taking their large-scale computing systems (TPUs, GPU-based systems, etc.) and making them available to everybody. Second is the software stacks (JAX, TensorFlow, scikit-learn, etc.). The third is more advanced solutions based on the requirements of the customers (AutoML, image, video and audio translation, etc.). Last but not least, the company has built a complete packaged solution.
“So, we feel that the boundary of what machine learning and AI can do will change over time,” said Kurian. For instance, he said, it was about doing assistive things when it started. Assistive things are what a human being can do, but the computer assists the human being in some ways to do it better.
This was followed by something you couldn’t do with humans because of the quantity of data you need to process or it might be far too significant. “So, the machine is doing something that humans couldn’t do, but it is still an incremental element on top of what humans could do themselves,” said Kurian.
“The third phase is generative AI,” said Kurian, enabling people to express themselves differently.