Sep 17, 2026 · 45 min · 13 segments
Season four kicks off with Paula Goldman, author of the new book, “Manage the Machine: How to Harness Human AI Collaboration at Work,” Paula is also Salesforce’s Chief Ethical and Humane Use Officer…
Paula GoldmanGuestReid BlackmanHostAnd in the first episode of the season, we're going to talk about what does good AI human interaction look like, especially in a business context.
And the reason we're talking about that today is because I'm speaking with Paula Goldman, who is the, big title, Chief Ethical and Humane Use Officer at Salesforce.
And more specifically, she's the author of a new book called Manage the Machine, How to Harness Human-AI Collaboration at Work.
Her general claim is that we're not thinking very clearly about what good human AI interaction looks like and we can do a lot better.
We talk about, for instance, the pipeline problem for having experts if junior people go away or tend to go away and we have AI working in their stead.
We talk about things like what does positive friction or good friction look like in an AI human interaction? But I want to focus on three major claims that we talk about throughout the conversation.
I just find these the most interesting and most worth saying, hey, pay attention to this part.
So the first is what Paula describes as the biggest management problem of our day with regards to how humans interact with AI.
That is to say, we think about humans standing between the outputs of an AI and the outcomes that result.
But as AI uses more and more information to come to its outputs or quote unquote decisions, It makes it very hard for a human to actually process and verify and validate all that information and then act in a responsible and professional way.
And so what Paula thinks is that we should be thinking less about a human in the loop and more about what she calls human at the helm.
Look, we're going to have, Paula thinks, sort of AI teammates that we interact with.
Eventually, you come out with products or a set of products or results, whatever you did with the agents.
Who's responsible? And it used to be, Paul says something like, look, at the end of the day, it's humans who are still responsible for that end product.
Doesn't this get really murky when maybe it's the case that an AI agent came up with something that was not quite right upstream of the conversation and interaction with humans and that gets adopted and then it gets amplified throughout and that could be a problem.
The third thing that I want to talk about is a point that Paula highlights a number of times, which is that she thinks that humans have a certain kind of intelligence.
It is definitely debatable, but let's just suppose that AI has a certain kind of intelligence.
But rather, since there are different kinds of intelligence, how we get those different kinds of intelligence to combine and work together in a fruitful way.
As for me and what I've been doing this past summer, a lot of talks and promoting of my book, The Ethical Nightmare Challenge.
And if you have any comments, trenchant criticism, et cetera, about the podcast or about the book, you can email me at em at reidblackman.com. All right, let's go talk to Paula.
Read the full transcript.
Create an account to read the whole episode, search across every transcript, and follow the shows you care about.
And in the first episode of the season, we're going to talk about what does good AI human interaction look like, especially in a business context.
And the reason we're talking about that today is because I'm speaking with Paula Goldman, who is the, big title, Chief Ethical and Humane Use Officer at Salesforce.
And more specifically, she's the author of a new book called Manage the Machine, How to Harness Human-AI Collaboration at Work.
Her general claim is that we're not thinking very clearly about what good human AI interaction looks like and we can do a lot better.
We talk about, for instance, the pipeline problem for having experts if junior people go away or tend to go away and we have AI working in their stead.
We talk about things like what does positive friction or good friction look like in an AI human interaction? But I want to focus on three major claims that we talk about throughout the conversation.
I just find these the most interesting and most worth saying, hey, pay attention to this part.
So the first is what Paula describes as the biggest management problem of our day with regards to how humans interact with AI.
That is to say, we think about humans standing between the outputs of an AI and the outcomes that result.
But as AI uses more and more information to come to its outputs or quote unquote decisions, It makes it very hard for a human to actually process and verify and validate all that information and then act in a responsible and professional way.
And so what Paula thinks is that we should be thinking less about a human in the loop and more about what she calls human at the helm.
Look, we're going to have, Paula thinks, sort of AI teammates that we interact with.
Eventually, you come out with products or a set of products or results, whatever you did with the agents.
Who's responsible? And it used to be, Paul says something like, look, at the end of the day, it's humans who are still responsible for that end product.
Doesn't this get really murky when maybe it's the case that an AI agent came up with something that was not quite right upstream of the conversation and interaction with humans and that gets adopted and then it gets amplified throughout and that could be a problem.
The third thing that I want to talk about is a point that Paula highlights a number of times, which is that she thinks that humans have a certain kind of intelligence.
It is definitely debatable, but let's just suppose that AI has a certain kind of intelligence.
But rather, since there are different kinds of intelligence, how we get those different kinds of intelligence to combine and work together in a fruitful way.
As for me and what I've been doing this past summer, a lot of talks and promoting of my book, The Ethical Nightmare Challenge.
And if you have any comments, trenchant criticism, et cetera, about the podcast or about the book, you can email me at em at reidblackman.com. All right, let's go talk to Paula.
