How do you apply generative AI in your role?
The way I'm already using it is by summarizing large pieces of content and deriving insights, different types of reports [and] transcripts of peer companies, for example, within a short period of time.
One thing to note is that I don't use this as a Google tool very much because that's when there's a higher risk of hallucination. That should improve over time, however. Currently, if I give it content and I know fundamentally that content is good and I focus it on summarizing or whatever the case may be, I have a much higher level of confidence that what I'm going to get back from the AI tool is going to be valuable.
How can generative AI and the IR role merge together?
If we think about generative AI in the same way we think about it being a human associate, we all have our strengths and opportunities for improvement. So when I build a team, I try to bring in people who complement existing strengths with additional skillsets and experiences. Generative AI is another associate but online. I’m going to use it to help make sure I’m thinking about things in a variety of ways (leveraging its persona capabilities, for example), or it might surface some new ideas.
I will probably also lean on it a bit more in areas I'm not innately good at. As an aside, I’m not much of a social media person so I don't use it much for that myself even though it can be quite good for those use-cases but, in any case, I think you have to understand where the opportunities are to help improve your function.
Can you explain whether generative AI will replace the necessity of an IRO?
I'd say that if IROs don’t, in my opinion, very soon start to tinker and experiment with generative AI, then their role could very well be diminished by someone who is willing to step in and leverage it as a tool to empower their capabilities. I just feel strongly that people should tinker now, safely and responsibly, because we know these things are not secure.
People shouldn't use non-public information, of course. That said we're already moving toward other things that will allow us to do some of that.
I believe strongly that it will not remove the IR profession, but it could very well have the possibility of transforming it significantly.
What would you want a generative AI regulation to do, if one was to come into effect globally?
That's a great question. I'd say maybe one thing could be to make it clear when you're using generative AI to generate a significant or material part of what you’re producing. There definitely must be guardrails around people using it nefariously.
I think it has to be a global framework, which is going to be very difficult in and of itself. But I’m hopeful that there's a consensus for generative AI because I do think it's going to be pervasive and, as a result, it should be regulated globally in a reasonable manner.
What else do you see in the use of generative AI for an IRO?
I guess one point I just wanted to make sure I made is I talked about ChatGPT a lot. I certainly use that more than anything else. But I'm also trying to look across all the tools; personally, I still find ChatGPT to be the best one, but there are others out there.
I think it's worth noting that people shouldn't think of generative AI as only ChatGPT. They should take a look at other tools to make sure they are using ones that offer the best function for what they need. There are new ones coming out regularly at this point. For example, Claude 2 by Anthropic seems to be one of the strongest challengers to ChatGPT 4 as of late.