Chatbot perspective from an insider (Rodney Brooks and Wired.com)

What Will Transformers Transform?

Thanks Rodney for a thoughtful discussion of

  • The Hype Cycle (peak of overinflated expectations)
  • The caution needed as our tools grow in skill exponentially
  • The ongoing risk of hallucination and unexpected errors in chatbots
  • The “grounding” problem with AI and robots

I particularly love the following quote:

Roy Amara, who died on the last day of 2007, was the president of a Palo Alto based think tank, the Institute for the future, and is credited with saying what is now known as Amara’s Law:

We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

I’m feeling an upward acceleration of AI skill with GPT applications, which could open up substantial risks as well as benefits.

For example, in the EHR space, there is discussion that GPT could take a patient and physician’s recorded conversation and automatically write a progress note, qualitatively more accurate and much faster than current commercial tools. Further, it could potentially summarize weeks of progress notes on hospitalized patients and write the discharge summary, a document that, when well-written, can take a human many hours of work. Or even, receive patients’ incoming MyChart messages and clinical questions and “reply to the patient in the voice of their clinician” based on the decade of writing by that clinician in the EHR.

Sure, these seem great. How about the potential deluge of GPT agents writing notes, requests ON BEHALF of patients, or other authors, that could junk up our systems? If 183,000 incoming patient messages is a lot (current monthly patient message volume at UCHealth), what if GPT somehow enabled 10x that number?

How much discussion will be GPT talking to GPT on behalf of employers/ patients/ discussants? I understand science fiction editors now have a 10x increase in sci-fi story submissions, a SUBSTANTIAL FRACTION now being written by GPT based on prior stories?

Too, wired.com has written about this:

https://www.wired.com/story/chatgpt-pause-ai-experiments-open-letter/

“I worry that we are very much in a ‘move fast and break things’ phase,” says Holstein, adding that the pace might be too quick for regulators to meaningfully keep up. “I like to think that we, in 2023, collectively, know better than this.”

CMIO’s take? Take a breath, everyone. It’s going to be a bumpy ride.

Author: CT Lin

CMIO, UCHealth (Colorado); Professor, University of Colorado School of Medicine

Leave a Reply

%d