Indisputable authorship ordering method:
Reviewer 2 should have blocked this for not specifying the version used (N64? Melee? Brawl? This is science!).
Hi! I’ve been wanting to revamp my site for years, but some mandatory security updates have forced my hand, so now we’re in the middle of frantic unplanned unavoidable total website redesign (oops)!
Please excuse any quirks as I work on this in my copious free time.
People have really enjoyed asking the current generation of large language models (LLMs) like ChatGPT about AI: a sort of overly on-the-nose meta-discussion. Reid Hoffman, co-founder of LinkedIn and techno-optimist/evangelist, co-wrote a free book about AI with GPT-4 called Impromptu: Amplifying our Humanity Through AI.
In a late chapter, Hoffman makes an argument that modern humans are homo techne (“the technological human”), essentially that “technological innovation is, and always has been, humanity’s defining feature.” He talks about the many inventions of the past and how amazing it has been since the early 1920s. For example:
Whoever you were in say, 2005, you are—if you regularly use a smartphone now—a dramatically enhanced version of that person. (p. 192)
GPT-4’s response is a bit more nuanced:
I don’t necessarily disagree with any of the points you made, but I think it’s important to remember that not everyone experiences these benefits equally. For example, while it’s true that the average human life has doubled in the past hundred years, there are still parts of the world where life expectancy is much lower, due to poverty, lack of access to healthcare, or other factors. And while smartphones and other technology have certainly made our lives easier in many ways, they can also exacerbate existing inequalities—for example, by making it easier for those who can afford them to access information and opportunities, while those who can’t are left further behind. So while I don’t disagree that we’re living in an incredible time, I think it’s important to keep in mind that these benefits aren’t universally shared.
Reid:
That’s a very important point. Do you think modern emerging technology can help us democratize these benefits, or otherwise address long-standing equity issues?
GPT-4:
I think it’s definitely possible. For example, telemedicine could help address the lack of access to healthcare in under-resourced areas, and technology that allows for remote work or online learning could create new opportunities for people in rural or impoverished areas. But I also think that it’s important to remember that technology is only a tool—it’s up to us as a society to make sure that it’s used in a way that promotes equity and benefits everyone, rather than just a select few.
There’s a certain grain of techno-optimism that’s almost as tedious as the doomsaying: the belief that good arises organically as a feature of the technology itself as opposed to from the human systems that deploy it.
For a lot of people, the early democratic internet seemed to bear this out. This is what I thought as a middle schooler playing online hand-coding websites in HTML to post on Geocities. The whole thing is magic! And then we got Wikipedia–amazing! But I think in the era of social media and conglomerate capitalism, we should all know better by now. Hoping everything gets so profoundly broken that the governments of the world will be forced to provide a universal basic income (UBI) but that otherwise no harm is done is probably naive.
GPT4 is derived from the content of the internet. While the model is tuned to provide more useful and less racist/sexist/terrible responses, the fact that its statistical process is able to generate responses like this suggests to me that the people who have written about this publically have probably thought about it correctly. I don’t know if we’ll listen.
When you look for the answers needed to confirm your beliefs, you can almost always find evidence. That doesn’t mean you’re right. It means confirmation bias is a real cognitive trap.
Radiologists (or clinicians of any stripe) need to constantly regulate and bring to consciousness balanced decision-making between observation and synthesis (putting together multiple findings to reach a conclusion) and anchoring on initial observations in ways that can impair objective analysis.
As in: is this additional imaging or clinical finding subtle or simply not there?
Imaging interpretation is a surprisingly noisy process. Sometimes we simply don’t know if a finding is “real” or not—we make judgment calls based on intuitive probabilities all the time. When findings make sense for a given clinical picture, we are more likely to believe them. Conversely, when we know what to look for, we are more likely to marshall our attention effectively and be able to identify subtle findings.
But: balance in all things.
There are two facets of confirmation bias that deserve their own discussion here: cherry picking and selective windowing.
Cherry Picking
You can’t retrospectively judge the likelihood of an event after the fact. This is part of the unfairness of Monday-morning quarterbacking and medical malpractice. You can’t predict the weather that occurred last week. Forecasting is a prospective process.
From Richard Feynman’s classic The Meaning of It All: Thoughts of a Citizen-Scientist:
“A lot of scientists don’t even appreciate this. In fact, the first time I got into an argument over this was when I was a graduate student at Princeton, and there was a guy in the psychology department who was running rat races. I mean, he has a T-shaped thing, and the rats go, and they go to the right, and the left, and so on. And it’s a general principle of psychologists that in these tests they arrange so that the odds that the things that happen by chance is small, in fact, less than one in twenty. That means that one in twenty of their laws is probably wrong. But the statistical ways of calculating the odds, like coin flipping if the rats were to go randomly right and left, are easy to work out.
This man had designed an experiment which would show something which I do not remember, if the rats always went to the right, let’s say. He had to do a great number of tests, because, of course, they could go to the right accidentally, so to get it down to one in twenty by odds, he had to do a number of them. And it’s hard to do, and he did his number. Then he found that it didn’t work. They went to the right, and they went to the left, and so on. And then he noticed, most remarkably, that they alternated, first right, then left, then right, then left. And then he ran to me, and he said, “Calculate the probability for me that they should alternate, so that I can see if it is less than one in twenty.” I said, “It probably is less than one in twenty, but it doesn’t count.”
He said, “Why?” I said, “Because it doesn’t make any sense to calculate after the event. You see, you found the peculiarity, and so you selected the peculiar case.”
The fact that the rat directions alternate suggests the possibility that rats alternate. If he wants to test this hypothesis, one in twenty, he cannot do it from the same data that gave him the clue. He must do another experiment all over again and then see if they alternate. He did, and it didn’t work.”
His conclusion?
“Never fool yourself, and remember that you are the easiest person to fool.”
This is also why when we evaluate a new AI tool, we don’t just judge how well it works on its training data. That information doesn’t help us predict how well it will work in the real world.
Cherry picking is seductive, which is why it’s so easy to fool yourself. We can’t just learn key lessons from post hoc judgments.
Selective Windowing
Selective windowing refers to the tendency to selectively seek and interpret the subset information that confirms our pre-existing beliefs or expectations while ignoring or discounting information that contradicts them. By analogy, a window constrains your view of the outside world.
The selective windowing of attention can dramatically skew decision-making.
I had an attending once who would review a case, and upon seeing one finding pointing in a direction, “see” several subtle supporting features to confirm a diagnosis. I assume some of this ability stemmed from experience and reflected true expertise.
But, some residents would also play a game during readout where they would describe the patient’s symptoms but purposefully not mention the side, and the attending would concoct a tidy narrative beautifully tying together a number of subtle observations. The problem, as I’m sure you guessed, is that frequently it would be the wrong side. The observations were only possible through that selective window. Too narrow a window and your view of the world is woefully incomplete and distorted. To torture another metaphor, the anchor of that initial observation sunk the proverbial diagnostic ship.
But, in practice, what a fine line to walk! Being sensitive to subtle manifestations of a complex process versus just seeing what you expect to see. Many radiologists have pet diagnoses that they call more than their colleagues. There are neuroradiologists who seem positively primed to see the findings of idiopathic intracranial hypertension or normal pressure hydrocephalus. Some of them are even assuredly better, more thoughtful radiologists. But some aren’t. Some will anchor on an initial observation and confirm their way to the story.
* * *
Attention is a finite resource. The world is too rich and vibrant to be seen unfiltered. We are always windowing, and when faced with important decisions, we must always seek to widen our window to consider competing information and address alternative explanations. Evidence is ubiquitous: it’s usually easy to find support for your preferred position, even when it’s wrong.
Delightful:
Monotony, the cardinal and most common sin of the public speaker, is not a transgression—it is rather a sin of omission, for it consists in living up to the confession of the Prayer Book: “We have left undone those things we ought to have done.”
Emerson says, “The virtue of art lies in detachment, in sequestering one object from the embarrassing variety.” That is just what the monotonous speaker fails to do—he does not detach one thought or phrase from another, they are all expressed in the same manner.
To tell you that your speech is monotonous may mean very little to you, so let us look at the nature—and the curse—of monotony in other spheres of life, then we shall appreciate more fully how it will blight an otherwise good speech.
If the Victrola in the adjoining apartment grinds out just three selections over and over again, it is pretty safe to assume that your neighbor has no other records. If a speaker uses only a few of his powers, it points very plainly to the fact that the rest of his powers are not developed. Monotony reveals our limitations.
In its effect on its victim, monotony is actually deadly—it will drive the bloom from the cheek and the lustre from the eye as quickly as sin, and often leads to viciousness. The worst punishment that human ingenuity has ever been able to invent is extreme monotony—solitary confinement. Lay a marble on the table and do nothing eighteen hours of the day but change that marble from one point to another and back again, and you will go insane if you continue long enough.
Project Gutenberg–a library of online/ebook public domain works–is amazing. Why else would you casually come across books published over a century ago like The Art of Public Speaking by Dale Carnagey and J. Berg Esenwein.
The final trailer for the Super Mario Bros. movie dropped, and it looks like it might be the rare (only?) (did you like the Sonic movie?) video game movie that hits the important game notes for enjoyable fan service while also looking like a decent movie.
Our family is genuinely looking forward to seeing this one.
[This post was originally written in September 2022. But, update: Yum. My third order just arrived in the mail yesterday.]
The coffee at work has been—for most locations of my training and attending career—terrible. From the burnt “Parks and Coffee” drip sitting for hours on the hot plate during residency to the cheapest K-cups at the imaging center. It doesn’t matter what sweetener or creamer you might add, it was rare to finish the cup once it was cold enough to taste. I’m not a coffee snob. I’m really more of a pragmatist. I don’t have the time or inclination for a long ritual even when working from home let alone the desire to do anything elaborate at work. Visiting the hospital Starbucks is slow and expensive. It’s a treat on the way for an early call shift morning but not something I enjoy doing routinely.
One of my residents shared Cometeer with me. It’s a variation on a coffee subscription (which is itself a variation on the incredible number of subscriptions available these days). I generally don’t like these sorts of things (who wants to remember to pause or cancel?), but I enjoyed the one I tried at work so I gave it a try.
The twist is that it’s a small recyclable cup of liquid-nitrogen flash-frozen coffee concentrate. Add your liquid of choice, hot to the frozen puck or cold after thawing, and you have instant coffee that isn’t, well, instant coffee. I predominantly use it as an espresso shot equivalent for making ice lattes, and for this purpose, it is effective and efficient. And it’s easy to slip in my bag and use on-site. On the enjoyment scale, I put it way above Starbucks’ regular iced coffee and just underneath their shaken iced espresso.
I honestly don’t know if I will continue to subscribe in the future, because it’s not the cheapest. You can cancel at any time and thankfully it’s also easy to spread out deliveries and pause them for months, because there’s no way I would want tons of coffee piling up in my freezer at any given time. I just don’t drink that much. With the new $20 new member coupon, the per cup cost is about $2 per pod (more than Nespresso, less than a coffee shop, and also more expensive than nothing, free tea, or the burnt brown caffeinated sludge otherwise available).
So, if you happen to be in the market for a new caffeine source and are interested in trying something new, you can try Cometeer and get $20 off (and subsidize my coffee consumption by the same amount!). Note that this is not a sponsored post; I just wouldn’t mind having cheaper coffee in the future. Also, note that I have literally never done a post like this in this site’s 13-year history. And finally, note that I can only use one referral bonus per order, so alas no matter how many of you choose to buy some I won’t be getting any coffee for free.
Lina Khan, in an op-ed for The New York Times:
Noncompete clauses systemically drive down wages, even for workers who aren’t bound by one. Every worker stuck in a job represents a position that isn’t opening up for someone else. And if employers know their workers can’t leave, they have less incentive to offer competitive pay and benefits, which puts downward pressure on wages for everyone.
F.T.C. economists conservatively estimate that noncompetes suppress American workers’ income by roughly 3 percent to 4 percent, or $250 billion to $296 billion.
Perhaps this would be more forgivable if noncompetes really did spur innovation, helping companies take big swings and bold risks. But here, too, the real-world evidence reviewed by the F.T.C. undermines the theory. As the name implies, noncompete clauses tend to make markets less competitive. Rather than encouraging dynamism and new ideas, they can enable stale incumbents to lock out new rivals.
Her essay is well worth reading. Khan makes a compelling case for the FTC’s proposed rule.
Competition is supposed to be a good thing.
From Nobel-laureate Daniel Kahneman’s Noise: A Flaw in Human Judgement:
More broadly, our sense of understanding the world depends on our extraordinary ability to construct narratives that explain the events we observe. The search for causes is almost always successful because causes can be drawn from an unlimited reservoir of facts and beliefs about the world.
[…]
This continuous causal interpretation of reality is how we “understand” the world. Our sense of understanding life as it unfolds consists of the steady flow of hindsight in the valley of the normal. This sense is fundamentally causal: new events, once known, eliminate alternatives, and the narrative leaves little room for uncertainty. As we know from classic research on hindsight, even when subjective uncertainty does exist for a while, memories of it are largely erased when the uncertainty is resolved.
Noise is a good book. It’s not an exciting book–though neither is Thinking, Fast and Slow for that matter–but it’s an important book. Bias is important, but human noisiness and our imperfect and very noisy systems are also important: Judgments (predictions, decisions) are hard to get right, and so much of the world’s punditry is just storytelling.
Doctors make dozens if not hundreds of little and big decisions every day. And I know that not only I am noisy, but I’m noisy in a variety of ways. A big part of my deliberate practice as it pertains to radiology is to try to limit how much noise distorts my decision-making and the quality of my reports.
We want to be not just good but consistently so.
I came across this brief article (“It’s Always Your Fault“) from 2016 by DHH, who–among other things–was the creator of the web application framework Ruby on Rails and co-founder of Basecamp/37Signals:
There’s a system in place that caused this to happen, and you’re part of that system. Shit never happens in a vacuum. The vast majority of it is a predictable consequence of the way things are. Even if it was “just somebody’s fault”, others put or kept that person there.
The goal is to change the system, and to change the system, you have to change its parts. Have the courage to start with yourself. Absorb as much blame and responsibility you can for what happened, and hopefully some of that introspection will rub off on the other parts of the system. But even if it doesn’t, you’ve still done your bit to improve matters.
In Medicine, we seem to oscillate between blame-game individual-at-fault finger-pointing and Just Culture the-system-is-the-problem.
It’s true we shouldn’t go around punishing people who are trying to learn and doing their best, and equally true that we need to always be looking to address system flaws. It’s also critical to keep in mind how many people working in healthcare are second victims of those mistakes, which prevents healthy introspection in favor of guilty misery.
But I also found his point just a little refreshing. As usual, it’s not either/or, it’s both.