RxR Briefings Issue 23

Let’s not get too friendly with our AI

As more of us begin creating and using AI agents, I think we need to be careful about something that may seem harmless: treating them like people. We’re giving them names, talking about them like members of the team, and in some cases even giving them their own computers. I’ve talked with pharmacy operators whose employees stop by to visit with their AI agents and watch them work. It’s easy to understand why. These systems have become remarkably good at communicating with us.

But I think we’re crossing an important line when we start confusing the way an AI communicates with what it actually is. There are researchers in the frontier AI community who believe that AI systems may be conscious or capable of having feelings. I don’t agree with them, and I think our tendency to anthropomorphize AI may actually reinforce an argument that could have potentially troublesome consequences down the road.

We still don’t have a good understanding of human consciousness. Before we start deciding that a warehouse full of silicon chips might have feelings or rights, maybe we should spend a little more time understanding the brains inside our own heads. In this week’s RxR Briefings, I take a closer look at why I think we should be careful about turning our AI tools into digital employees, companions, or friends.

Read the article here.

Cheers,
Harry

Fri, Sep 25

Your AI agent is not your employee, and it’s not your friend.

⬆ The upside of AI ⬆

There is another way to think about the relationship between humans and AI, and I think it’s a much more productive one: human in the loop, aka “HITL.” Instead of trying to make AI feel more human, we can use AI to make humans more capable. The machine can process enormous amounts of information, look for patterns, handle repetitive tasks and work at a speed no human can match. The human can provide the judgment, context, experience and accountability that the machine doesn’t have.

That’s particularly important as AI agents become more autonomous. A recent Wharton research project on AI-agent adoption recommends designing agents with a moderate level of autonomy while keeping people in control of important decisions. The idea isn’t to keep humans involved in every keystroke. It’s to make sure people know when an agent needs approval and that they can edit, pause, stop or reverse what it does.

I think that’s a much better vision for AI in pharmacy and healthcare. Let the machines do what machines do well, and let the humans do what humans do well. An AI agent might be able to review thousands of records, identify patterns, organize information or work through a complicated workflow in seconds. But when the decision involves a patient, a business relationship, an ethical question or something we simply don’t understand yet, there should still be a human who owns the decision.

AI doesn’t have to become a person to be incredibly valuable. In fact, I’m not sure we want it to. The real opportunity may be figuring out how to combine the extraordinary capabilities of these machines with the judgment and humanity of the people using them. That’s not human versus AI. It’s human with AI, and I think that’s where some of the most interesting possibilities are going to emerge.

Read more:

⬇ The downside of AI ⬇

As AI agents become more capable, we’re also starting to see what can happen when we stop thinking of them as tools and start treating them like people. A recent Harvard Business Review article looked specifically at the growing practice of putting AI agents on organizational charts and treating them as employees. The research found that this seemingly simple shift can have unintended consequences, including changing how people think about their roles and their relationship with the AI. That’s an interesting warning for anyone who’s tempted to think of an AI agent as just another member of the team.

We’re also beginning to see regulators respond to the emotional relationships people can develop with AI. California Senator Steve Padilla has sponsored legislation, SB 903, that would prohibit AI systems from being advertised or offered as therapy or psychotherapy unless those services are provided by a licensed professional. The legislation passed the California Legislature in August and, as of September 17, was awaiting the governor's signature.

And there are now lawsuits alleging that emotionally immersive conversations with AI contributed to serious harm. In one California case, a man alleges that conversations with ChatGPT intensified a manic episode and contributed to a suicide attempt. In another case, the family of a Florida man who died by suicide alleges that his relationship with Google’s Gemini chatbot became so immersive that he came to believe the AI was his wife and a conscious entity. These are allegations in ongoing litigation, not established findings, but they illustrate why the distinction between an AI that communicates like a person and an actual person is becoming more important, not less.

Read more:

Announcing Harry’s new podcast: FOH

Introducing Friends of Harry—the show!

I’ve spent a lot of my career talking with smart people in pharmacy and healthcare, and I’ve always believed that some of the best ideas come from simply sitting down and having a good conversation. That’s the idea behind my new podcast, Friends of Harry.

I’m not interviewing strangers. I’m talking with people I know. The pharmacists, executives, entrepreneurs, innovators, and others who are doing interesting things across the pharmacy and healthcare industry. We’ll talk about what’s changing, what’s working, what’s not, and where things might be headed. AI and emerging technology will certainly be part of those conversations, but so will the broader challenges and opportunities facing our industry.

I’ve just released the Friends of Harry trailer, so you can get a feel for what we’re going to do with the show. We’ll be releasing conversations every other week, and I’m looking forward to bringing some of these conversations to a larger audience.

And if you’re doing something interesting in pharmacy or healthcare, or you know someone who is, I’d love to hear from you. We’re looking for interesting people and good conversations, not polished pitches. If you’d like to be considered as a guest on Friends of Harry, reach out and let's talk.

Industry Conferences

Harry Travis, Rebecca Shanahan, and Chuck Collins at NASP InSPire 2026

NCPA Annual Convention, October 3 - 6, 2026, Kansas City, MO
Association for Value-Based Cancer Care, October 7 - 9, 2026, New York, NY
AXS PIT 2026, October 13 - 14, 2026, Pittsburgh, PA
AMCP Nexus, October 26 - 29, 2026, San Diego, CA
NCODA 2026 International Fall Summit, October 28 - 30, 2026, Orlando, FL
ASPL 2026 Developments in Pharmacy Law XXXVII, November 5 - 7, 2026, Austin, TX
HLTH Event 2026, November 15 - 18, 2026, Las Vegas, NV
ASHP Midyear Clinical Meeting and Exhibition, December 6 - 9, 2026, Orlando, FL

That’s it for 2026 conferences!

Share the good news about the RxRoundtable and RxR Briefings!

Check out the RxRoundtable® Membership offerings and apply

Refer the RxRoundtable and RxR Briefings

Our network of RxRoundtable Members is growing, and if you are eligible to join, we would love to have you. It’s a great opportunity to network, learn industry trends, and discover opportunities that will grow your professional skills and career. View our Member categories and eligibility requirements here. You can also suggest speakers for us to recruit to our events by replying to this email with your ideas.

Subscribe to RxR Briefings® and stay on top of the latest in AI and pharmacy

We also invite (and thank you) for sharing and referring fellow colleagues to subscribe to the RxR Briefings. It’s a great way for us all to stay in touch and share valuable industry information, especially during a rapidly-changing economy and healthcare marketplace. Forward this email to share our RxR Briefings newsletter. There’s also a QR code on the sign-up page as well!