Are we truly entering the AI era?
Infosys certainly thinks so. And is going with all guns blazing on the AI-first bandwagon.
We just returned from the Infosys annual customer conference, Americas Confluence 2023, in San Diego, on the theme of AI-first everything.
Now, this is hardly surprising as in the last 10 months, all my conversations seem to be related to generative AI (Gen AI). Whether I am speaking to enterprises, service providers, influencers, or even with my six-year-old daughter, every conversation has elements of Gen AI (well, maybe not ALL the conversations with my daughter. Some of them have been about ice creams and toys, too). No other technology has taken our industry by storm as much as Gen AI. Consequently, every conference, seminar, or webinar has been about this technology and all the hyperbole around it.
So, I was mentally prepared for an avalanche of Gen AI at Confluence. What I wasn’t prepared for was Infosys’ unique and pragmatic approach to Gen AI. They are clearly done with all the hype and hoopla and are now focusing on the nuts and bolts of bringing the revolution to fruition for their enterprise customers.
Over the two-day conference, two messages emerged quite clearly:
-
- AI is not a silver bullet for everything, at least not at the current level of maturity.
- The real opportunity lies in the plumbing for AI—data and cloud.
The second session, by renowned AI rationalist Gary Marcus, set the tone as he talked about the various shortcomings of AI as it exists today. The session focused on setting a realistic tone for the proceedings by discussing examples of hallucinations and other problems and how even the leading lights of the AI revolution are advocating caution. The writing on the wall was clear—we are nowhere near the maturity that enterprises need for consistent results. There is much more work to be done, both in the core technology and in setting realistic expectations and understanding the limitations of current tools.
This view aligns very well with Avasant’s position on Gen AI. All new technologies undergo periods of hyperbolic expectations (remember blockchain in 2016–2017?). Still, the sooner we reach realistic expectations, progress can be made toward identifying use cases that will truly benefit from Gen AI. In fact, we have taken a pragmatic look at Gen AI in our recent publication, Generative AI Idea to Action: Execution Strategies for Enterprises and Service Providers, where we’ve talked a couple of levels below the hype and covered emerging deal constructs and pricing models. We also covered the real productivity gains from Gen AI that will underpin these new models. Several of our Gen AI research reports have discussed the real-world challenges of adopting the technology.
This brings us to the second message, which is most important for both enterprises and service providers. As we track this space, new models or advances in existing models are being announced nearly daily. It is easy to get scattered in one’s approach as one tries to stay abreast with all the changes and figure out what every new model means for one’s business (both for enterprises and service providers). However, Nandan Nilekani, Infosys’ co-founder and chairman, gave a very interesting and valuable mantra:
“Models are transient, data is persistent.”
This gives the North Star to Infosys in terms of what part of the problem they should be attacking. Irrespective of the AI model used, only those enterprises that are cloud and data-ready will be able to take true advantage of Gen AI. Infosys wants to become the provider of choice for that part of the problem of developing and implementing the data and cloud foundations at scale for enterprises to enable their journeys toward becoming AI-first.
As a part of this strategy, Infosys’ AI-first suite of offerings, clubbed under the Infosys Topaz moniker, brings in elements of Infosys Cobalt cloud offerings and their extensive data analytics capabilities. Using Topaz, Infosys is already working on 80 exploratory Gen AI projects and expects four or five to soon reach the pilot and monetization phase. To make things even easier for enterprises, they have pivoted many of their AI offerings along the industry lens and consequently have focused AI capabilities on all major industries. Specifically, Infosys is building capabilities that span horizontal custom AI apps on closed and open models to industry-specific AI apps using specialized pre-trained models.
Another major issue concerning AI adoption is data privacy and cybersecurity. As we have been saying in all our cybersecurity publications, security cannot be an afterthought. Systems need to be developed where security is a day-one action item. Most providers we meet with have been talking about this, but only a handful have been able to demonstrate client case studies where this is executed. So, it was heartening to hear that Infosys is focusing on AI solutions that are security and privacy first by design. It is developing foundational model moderation APIs that ensure both inputs and outputs are subject to security, privacy, and usability checks. The company has also established an AI risk council to review use cases and assign requisite moderation checks. It also showcased real-life examples with active clients where it implemented this approach.
AI has captured the world’s imagination in 2023. We are at the beginning of what looks likely to become the age of AI. However, there is a long journey with several challenges to overcome. The core technology needs to become more consistent and integrate with other fundamental technologies, such as quantum computing. But long-term benefits of AI will only be achieved by enterprises that invest today in building their data infrastructure across data generation and management, cloudification, data privacy, and cybersecurity. Service providers are well-positioned to guide and support enterprises in building this infrastructure and become trusted partners on enterprise AI journeys.
Akshay Khanna, Managing Partner; Shevy Magen, Partner; and Swapnil Bhatnagar, Sr. Research Director, attended Americas Confluence 2023 at Infosys’ invitation.
