Selling Probabilities
Updated: Jul 25, 2024
As a Tech Leader who is building AI products,
One of your major hurdles is convincing your clients that your product is credible
Even though everyone calls it Artificial Intelligence, you know it is actually Artificial Inference.
That is because the output is always an inference.
A mishmash of probabilities, likelihoods, affinities, propensities and confidence intervals.
But your client doesn’t like any of those words.
They prefer certainty, not likelihood of an outcome.
Because probabilities make everybody nervous.
So your client will demand a reasonable amount of certainty from the output of your product.
So what does a leader like you got to do in this situation?
Here are a few tips:
Set realistic expectations with your client about what your AI product can and cannot do. They won’t like it, but this is the first step in helping the client adopt and improve the product.
Always refer to your AI product as an inference product., as a modeled output, as an inferred data product. This will trigger the client to think about the implications, and it will be an opportunity for you to coach them.
If you are building a data product, and if it also includes additional ‘AI Features’, then try to keep these two separate.
As a builder of new AI tech, the burden of helping users adopt this tech is on you, whether you like it or not. Adopt the attitude that you don’t just build the tech, but coach the community on how to use it,and be an evangelist.
When asked to explain the AI product stay away from difficult to understand algorithms, and away from machine learning jargon.
Users also tend to say that the AI product is accurate because it is ‘scientific’. There are however, no ‘scientific’ silver bullets . Discourage such generalizations.
Instead create examples and demonstrate use cases from a user perspective. One doesn’t need to know how the internal combustion engine works in order to drive a car. Establish what the product can and cannot do. Discourage the notion that it is ‘magic’.
For example, let’s say you built a simple AI chatbot to handle discussions about baseball. But if your client starts having conversations with the bot about soccer, it may not be constructive. You and your client both need to get comfortable with the boundaries of your inference product. Once the client is on board, they will champion your AI product.
Recent Posts
See AllIf you are an individual contributor, especially in a tech sector, sooner or later, you will think about whether you should move into...
The best way to think about AI is ... Think of its output as an opinion. And think of data as facts. And be very careful not to confuse...
If you aspire to be an AI Tech Leader...you need to adopt a very different mindset to be successful at it. If you are an aspiring AI Tech...