Sold with AI - Edition Three (Hype Around AI)

If there is one thing true about AI, especially Generative AI, it is the fact that it's overhyped. This edition focuses on diagnosing this hype in a little more detail.

When it comes to AI, one of the prominent topics on my mind is - what is real and what is hype? What are the use cases where we should be leveraging it right away vs. the use cases that we should be wary about?

You can’t fully trust the vendors of AI solutions (and we are one!) to guide you, because the vendors are builders - their mindset is to build and try. So in that quest, some of them behave responsibly, but many do not.

So today, I will talk about the hype associated with AI, and what one should possibly stay away from. We will focus especially on Generative AI (which is a good time to remind that AI does not mean ‘just’ Generative AI)

Hype around AI, especially Generative AI

If you ask Gartner, they would tell you that they expect 60% of Seller Work to Be Executed by Generative AI Technologies. Within the next five years!

Now that’s quite a number, and that’s quite a statement. If more than half of sellers’ work is going to be executed by AI, then does that mean that we should expect seller productivity to double by 2028? It’s not impossible.

On the other side, you have the AI pessimists (they might prefer to call themselves realists). Eric Siegel, a seasoned academic in Machine Learning, writes in Harvard Business Review that an AI winter is imminent in the next 5 years. Forrester seems to be taking a somewhat more guarded approach too.

But the one I pay more attention to (mainly because they have a practical POV, not academic), strangely enough, is this VC fund called Emergence Capital. They say that today’s Gen AI is not tuned for B2B. Now, arguably, this article was written around a year ago and things have been changing fast, but I do think there is merit to their points. You should give it a read. 

Generative AI is definitely in peak hype mode right now, that is clear to everyone. But there are some needle-moving possibilities emerging, and it would be foolish to ignore them as well.

Let us dive into some specifics.

Overall, when it comes to Sales, I do not think that we are ready for autonomous deployment of Gen AI yet - where AI does things by itself, without human oversight. There are startups that are beginning to offer such solutions but I don’t think AI is ready yet. (We ourselves are still in wait and watch mode right now)

Let me take an example. We have been working with Open AI for close to two years now. Amongst other things, we ask it to rewrite existing email templates in a personalized manner (based on what our own AI has predicted about that person). In spite of multiple prompt iterations, sometimes it still starts answering questions in an email instead of rephrasing them. And our engineers say that  it happens more often when a user does not accept a first output, and asks it to rephrase the same question again. In its quest to provide a ‘different’ answer that we may accept, the AI starts trying different approaches.

There are many such examples.

Another area that gets talked about often is data management/automation. Salespeople have historically been bad at manually entering data into the CRM. While many products have tried to automate it, CRMs are still in a far-from-perfect shape.

I believe that AI is ready enough for automating unstructured notes. Summary of a meeting for example. But can you automate structured, or more specific data? For example, various teams who have a say in decision making at this particular customer. Or the exact costing that is likely to get approved vs. not. Yes, AI can learn from past deals, and it can infer the reactions to your proposal, but knowing the specifics of this given customer’s organization, or the value that is possible for them, is still hard.

And the final area where I believe AI’s potential is overhyped is forecasting, especially 1. long-term forecasting, and  2. early deal-specific forecasting. As deals get activated and more data flows in, forecasting becomes easier (and accurate). But long-term forecasting, when the deals are not yet activated, especially in markets that are not fully mature or companies that are still in growth mode is extremely hard to do. I doubt anyone was able to forecast NVidia’s growth spurt for example, simply because of the nature of change at play in the markets.

While these might be some areas where AI, especially Generative AI, could be overhyped, there are plenty of areas in sales where it presents a real opportunity. Today. 

In the next edition, I will talk about the opportunity areas.