The Promise of Generative AI
Earlier this week, I had the pleasure of facilitating an exchange between several Chief Information Officers of large Life Sciences Companies. The topic was Generative AI. It was a rich exchange of ideas among many leaders who have been planning how to make Generative AI solutions come to life in their respective organizations.
Gen AI is exciting for everyone present, and its promise is only beginning to be understood. In other words, it is not simply a fad — it is here to stay. “Generative AI will be the most disruptive technology that any of us will see in our careers.”
Gen AI is different and feels different than past false starts such as Metaverse, partly because it is so accessible and technologies such as GPT have been made consumer friendly. Participants anticipated impacts including higher levels of productivity, greater innovation, and transformed processes. “There is excitement for cost savings; gaining speed of operations; streamlining processes; all in all, for huge productivity gains.”
The Role of CIOs
Everyone has begun thinking about how to realize these benefits and most have an action plan. CIOs of most organizations will be the ones leading the charge. “We came up with our first architecture and put a sandbox in place… IT has taken the lead with generative AI. There’s clearly a strong business relationship, but given the automation and security, it must be us leading it. We are driving the scale.”
The participants in this Council all work in a regulated industry. And in regulatory environments, along with excitement, there is some caution — especially as it relates to the truthfulness, informativeness and propensity to hallucinate of these LLMs. “We put guardrails up and work closely with privacy and legal, but we also allow people to try stuff. We need to be careful we don’t stifle experimenting and learning so much that we don’t progress.” Companies are wrestling with ways to control risk without constraining innovation. There is risk in how these models will be used and deployed and, equally importantly, which data is being used to train them.
Augmenting Classical AI Approaches
Most solutions being considered are those with humans in the loop. Automated solutions are mostly reserved for instances where being wrong is not as big of an issue — Enterprise Search is the perfect example. “How do you keep humans in the loop to ensure accuracy, applicability of use, ensure data is clean, and we get a good outcome?”
Participants also posed a few questions on how Generative AI fits with the more classical AI approaches. Does Gen AI replace them, or merely augment them? Is it dependent on the nature of the problem being solved Infer, Predict, Generate Content) or the nature of the data that is being used (Structured vs. Unstructured)? “Generative AI will make us all super stupid (and we are unable to do anything about it).”
The Future of Generative AI
But the allure of productivity is strong. “It’s as transformative as the calculator was for personal productivity. In 10 years, those that take advantage of generative AI will likely have a 10x multiplier around their productivity.” For those that have not had AI maturity across multiple BUs within the organization, the consideration is, “Can we use this to leapfrog if our maturity is currently low compared to industry standards?”
And finally, there was a shared realization that Gen AI is likely not a cheap endeavor. “Sustaining cost-efficiency in Generative AI tool / platform decisions as the space matures and is monetized is a concern.” But the excitement of what the future may hold overwhelms the uncertainty. “It is intriguing and scary to think of what will soon be possible with Generative AI. I feel like a kid in a toy store.”
