Turning Points in AI and ML

We know what’s making waves today. Where is artificial intelligence and machine learning headed next?

Manoj Suvarna, Managing Director, AI Ecosystems, Deloitte

January 9, 2023

4 Min Read
abstract of AI and ML
ktsdesign via Adobe Stock

AI and ML are well-established as technologies that work, but still in their early days as technologies that get work done. Somewhere between curiosities and everyday workhorses, they’re evolving quickly, and the focus today is on applying them in ways that generate measurable business value.

This is where “what if?” meets “what’s next.” As with previous technology revolutions, the future rollout of these and other AI/ML applications will have less to do with experimental results in search of applications, and more with established business needs that the new tools can address in transformative ways.

For example: Thanks to the rise of MLOps, a company that merely “uses” ML today may regard it as a bedrock operating principle tomorrow, the conversational AI that lets today’s bots train to address a defined topic will be completely unshackled tomorrow, able to follow and contribute to any conversation on any subject. Instead of scanning retail products and monitoring traffic, computer vision may become a bulwark of social fairness and economic growth.

From Assistant to Decision-Maker

When one couples the technology capabilities of AI/ML solutions with the practical ends they may evolve to serve, the potential examples multiply quickly. Think about the job of content moderator in a social media environment, which requires not only discrete judgment, but also attention to detail at tremendous speed and volume. Today, bots can screen text for certain terms and expressions. But tomorrow’s may be able to monitor sentiment, understand contexts outside the immediate content, or evaluate non-text expressions, like videos or photos.

In a contact center environment, you can “trip up” a present-day bot by trying to stray from its trained area of knowledge. That’s because today’s conversational bots learn from other conversations. Tomorrow’s will learn from the world of unstructured inputs around them, as people do, so they’ll keep up with you no matter where you steer the conversation. Natural language understanding and large language models will translate that into unfettered interaction and follow up with required action that’s instantly and automatically informed by relevant data wherever it resides.

That means that while a conventional bot can help direct you to other resources when you’re discussing insurance policies, zoning laws, or other complex matters, tomorrow’s AI will be able to handle entire conversations and relationships.

The Entire Visual World as Data Source

Then there’s computer vision. We may sell that technology short when we think only in terms of live cameras trained on live processes. What if you had to find a specific visual element in hundreds of hours of online video? You could burn all those hours in real time -- or an AI-powered system could sift through it in a way that mimics the way a human searches memory. In other uses, the stakes will likely be higher: Safety on the highway, the tarmac, or the assembly line will rely on this technology as well. Your supermarket bar code scanner will still work, but it will have many more powerful cousins.

Sometimes these technologies gain power in combination. Think of a time when you’ve had to enter personal information into an institution’s registration system -- a health care provider, a travel company, or a bank. Conversational AI can take some of the burden off that process. Computer vision can contribute as well, for example by scanning the details from a driver’s license, or uploading a relevant photo, while the call is happening. One insurance company trimmed its sign-up process from days to minutes that way, via an experience that drew high satisfaction marks from consumers.

Bringing It Together

What will it take to realize visions like these? It’s time to think bigger. For example, computer vision has fast-growing power, but the data archival and storage architecture it takes to support it have to grow just as fast. Each of these technologies is surrounded by enabling capacities that need to grow alongside it.

It’s also time to hold these tools to a higher standard. Like anything else a business invests in, they need to show a predictable, measurable return. Their capabilities are just now passing the threshold at which that’s achievable in more applications, at greater scale.

What’s possible if we get there? Leaps past what we’re already used to. Consider that an AI system that uses technology to emulate all the human “senses,” in combination with decision-making ability, can fully function as a front-line worker in many situations. The next generation of AI isn’t task-specific but generalized in a way that will let it free entire cadres of human workers for more sophisticated and satisfying tasks.

We’ve heard about AI and ML for a while now. Most people have had some interactions with versions of these technologies -- ones we’ll someday think of as very basic. And their future evolution will follow a long path we can’t entirely predict. What’s clear right now is that they are at a turning point. They’ve proven they can work -- now it’s time for them to prove their worth.

About the Author(s)

Manoj Suvarna

Managing Director, AI Ecosystems, Deloitte

Manoj is a managing director for AI Ecosystems within the Technology Strategy & Partnerships team. In this role, he represents Deloitte’s AI practice in, identifying & executing ecosystem relationships with technology & data partners to support AI business growth initiatives.

Manoj started his career with Deloitte in early 2021. Prior to this he spent almost 24 years at Hewlett Packard Enterprise (HPE) in a variety of leadership roles spanning Product Management, Sales GTM, P&L Management & Alliances Marketing.

Manoj holds an MBA in Marketing from University of Memphis and a BS in Production Engineering from University of Bombay. He also completed an Executive Course on AI at MIT. He serves as a member of the AI Advisory Council @CompTIA and is based in Houston, TX.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights