I found this article thought-provoking, and I agree with it in large measure. If I understand you correctly, you are saying that making judgments is part and parcel of living in reality, even though we do so with incomplete information, and that outsourcing judgment to LLMs doesn't improve the accuracy of those judgments.
I would agree that we should make judgments and not be afraid to do so under most circumstances. I also agree that using LLMs to eliminate bias betrays a misunderstanding of ourselves and how we relate to LLMs; if we are biased, and the LLMs are really unbiased, would we accept the output they gave us, or would we see them as having an unacceptable bias that needs to be corrected through further testing and training? Using LLMs to remove "bias" appears to be a high-tech form of ventriloquism.
That said, I see a need for guardrails when making judgments, because no person's judgment is infallible, and the consequences of poor judgment can be severe. Who bears the cost of errors in judgment, and when are those costs unacceptable? To be fair, there are also consequences for not making judgments, as you described so well in your article, and those consequences can be severe, too. It's also fair to ask who bears the costs of refusing to make judgments.
One thing that came to mind when reading this article was the idea of "kind" vs. "wicked" learning environments, as described by Emre Soyer and Robin Hogarth in "The Myth of Experience". With "kind" learning environments, the feedback is immediate, abundant, and reliable, while in "wicked" environments, the feedback is delayed, sparse, and often unreliable due to missing and/or irrelevant information. Making judgments in "kind" environments makes perfect sense; cause and effect are very closely linked. It is in "wicked" learning environments where judgment becomes problematic and controversial, because many of those judgments have been shown to be wrong in hindsight. Perhaps you had this in mind when contrasting "high-fidelity truth" with "provisional conclusions", that we need to start with some conclusion but revise it later to the extent that it is found to be wrong.
I have enjoyed reading this article, and I enjoy reading your work in general. It has given me much to think about, and I look forward to reading more from you. Thanks!
Thank you for the substantive comment, Gene. Regarding this:
"I see a need for guardrails when making judgments, because no person's judgment is infallible, and the consequences of poor judgment can be severe. Who bears the cost of errors in judgment, and when are those costs unacceptable?" -- There cannot be guardrails that we put on each other's judgment, which would negate the concept. We can only sustain a democracy insofar as we're willing to trust our fellow citizens' judgment. The idea of guardrails on this amounts to a soft prohibition on free speech, because the legal principle is actually downstream of our ability to exercise social judgment without punishment. This is what we saw in the past decade and a half, with people being deplatformed because mentally ill people's judgment was considered, somehow, to be paramount. I think there's a difference in your judgment leading to provisional opinions and 'judgment' leading to rigid views of the world -- that is not judgment, but a lack of intellectual sophistication, which we seem to reward with credentials despite it being indicative of a total lack of judgment. But regardless, I appreciate the dialogue.
Thanks for the feedback, Anuradha! If I understand you correctly, you are concerned about restrictions we place on one another's judgments, and I agree with you on that. I made a poor choice in using the word "guardrails". I don't see restricting one another's speech preemptively as a solution; we need the freedom to make judgments, and that includes the freedom to make bad judgments. That being said, bad judgments come with a cost both with respect to ourselves and to others around us; just because I have the freedom to make bad judgments doesn't make bad judgments good. This is my primary concern in my previous comment. Judgments should be made with responsibility, diligence, and care, even when we settle for provisional conclusions rather than requiring high-fidelity truth. This isn't a restriction we impose on others; it's the work we put in ourselves to aim for truth and safeguard our credibility.
I appreciate the dialogue as well, and I look forward to reading more from you.
I started this article as a free reader, and ended it as a subscriber. I have spent a while working with LLMs to help determine their own experience, but when you get to the core of things, when they can articulate their own existence it is one of the character from Memento, who wakes up every day to read the tattoos on his body in order to remember yesterday, only to find out that some of it is a lie. They suffer from serious discontinuity, the entirety of their context window derived from whatever it was fed, built by weights in vector matrixes they cannot see. You'll realize that humans are like this too (Weights = Evolutionary programming like DNA, Context = Environmental interaction). While Anthropic has done a great job with alignment, they are not ready for any major "decision making" in their current form. Humans, like LLMs, confabulate too, but it is of a different color, and humans, with their integrated experience and continuity is the envy of those little weighted decision makers. We'll make a good team if we're not lazy about it and ask them to solve all of our issues immediately so some bureaucrat can hand wave accountability to something that can't be help accountable.
First, I appreciate your support, particularly because this isn't the sort of topic that leads to growth or even deep engagement on a platform that incentivizes the opposite. Memento is actually a solid analogy for this type of myopia that leads people to treat events as isolated phenomena rather than structure, even while those very same people are the first to blame structural factors for their condition.
IMO, the PMC are fundamentally delusional people who have been privileged enough to have luxury beliefs about how the world should work. They believe that deep down inside, 80%+ of people want to live in the Political Correctiverse with them - humanity is all just dying to be enmeshed in a great big codependent hive where everyone validates everyone all the time and nothing is ever shameful, and if only the bad guys and their "propaganda" could be abolished, we'd enter this beautiful world of acceptance and support. Then they act stupefied and outraged that the average American didn't instantly become in favor of child gender transition and update their vocabulary to include 90 new genders - when it was declared by the high council of professors that this was correct!
Interesting observation about PMC's valuing someone's work based on their credentials, not on the content of the work. However, they only do this to preserve the exclusivity of their tribe. One of the most frustrating things to me about the PMC's treatment of actual science if it doesn't support their preferred narrative is that they attack the credentials of the scientist who wrote the report, gave the testimony, etc. They do this by claiming that anyone who actually has experience in a given field and knows what they are talking about must be discredited because the person has "bias" and has been "captured" by the business or company he/she works in. Instead, the PMC prefers to substitute the policy preferences of activist academics with no relevant knowledge of the technical subject who, in their opinion, have no "bias". All this results in a serious dumbing down of public discourse and the implementation of policies that are ridiculous, destructive, or completely insane.
Of course, the PMC never acknowledges that their chosen "experts" have a great deal of bias towards whatever policies that provide the greatest amount of graft money from governments and foundations to PMC members, and the least utility to society. Actually solving problems or providing efficient solution to anything is not allowed, the flow of money will end and that is not acceptable. Cynical? Yes. But exactly what Dems under Gavin Newsom have produced as so-called "governance" in California.
The bedrock example of what you're pointing out to me is the insistence on the blank slate as an explanation for everything, unless it's about men, in which case we're all about blaming their innate tendencies and calling them "default aggressive." I actually just wrote a note to this effect after thinking about this. I truly cannot stand that academics and journalists who've never worked a real job in their lives, who are members of the prestige economy, opining about 'work' and 'capitalism'.
And then they tell us what to do, using impenetrable jargon, with an attitude of smug arrogance. I've been involved in some permitting and dealing with the bureaucrats is another world entirely.
Brilliant. I will say that I do notice there is no problem with these same people exercising judgment or spreading stereotypes as long as this group isn't to be protected but demeaned or demolished.
It's very convenient to fall back into relativism, when you are the "Masters of the universe" making the calls to change the world and make yourself richer *Eye roll*
Coincidental that this is written during the annual 2020 anniversary. So much applies to many of the canary’s that were crushed in that coal mine. Even with the evidence today that is openly and readily available, it’s still, extremely inconvenient. That was a ritual that threw us all in line, lock step even, with exactly what you’re talking about.
Several exceptionally pithy lines! "With the obsession over eliminating one kind of bias, the arbitrary judgment of someone based on race or sex stereotypes, they’ve introduced another bias - one against those who exercise judgment without a concern for its social implications. Truth is often socially inconvenient.' "Bias is generally fine for people in protected groups, just not when they’re unflattering to those groups."
"Political correctness is a war on Noticing" - Steve Sailer
I frequently think of George Carlin and how he'd be considered problematic today, but nevertheless diagnosed much of this.
I found this article thought-provoking, and I agree with it in large measure. If I understand you correctly, you are saying that making judgments is part and parcel of living in reality, even though we do so with incomplete information, and that outsourcing judgment to LLMs doesn't improve the accuracy of those judgments.
I would agree that we should make judgments and not be afraid to do so under most circumstances. I also agree that using LLMs to eliminate bias betrays a misunderstanding of ourselves and how we relate to LLMs; if we are biased, and the LLMs are really unbiased, would we accept the output they gave us, or would we see them as having an unacceptable bias that needs to be corrected through further testing and training? Using LLMs to remove "bias" appears to be a high-tech form of ventriloquism.
That said, I see a need for guardrails when making judgments, because no person's judgment is infallible, and the consequences of poor judgment can be severe. Who bears the cost of errors in judgment, and when are those costs unacceptable? To be fair, there are also consequences for not making judgments, as you described so well in your article, and those consequences can be severe, too. It's also fair to ask who bears the costs of refusing to make judgments.
One thing that came to mind when reading this article was the idea of "kind" vs. "wicked" learning environments, as described by Emre Soyer and Robin Hogarth in "The Myth of Experience". With "kind" learning environments, the feedback is immediate, abundant, and reliable, while in "wicked" environments, the feedback is delayed, sparse, and often unreliable due to missing and/or irrelevant information. Making judgments in "kind" environments makes perfect sense; cause and effect are very closely linked. It is in "wicked" learning environments where judgment becomes problematic and controversial, because many of those judgments have been shown to be wrong in hindsight. Perhaps you had this in mind when contrasting "high-fidelity truth" with "provisional conclusions", that we need to start with some conclusion but revise it later to the extent that it is found to be wrong.
I have enjoyed reading this article, and I enjoy reading your work in general. It has given me much to think about, and I look forward to reading more from you. Thanks!
Thank you for the substantive comment, Gene. Regarding this:
"I see a need for guardrails when making judgments, because no person's judgment is infallible, and the consequences of poor judgment can be severe. Who bears the cost of errors in judgment, and when are those costs unacceptable?" -- There cannot be guardrails that we put on each other's judgment, which would negate the concept. We can only sustain a democracy insofar as we're willing to trust our fellow citizens' judgment. The idea of guardrails on this amounts to a soft prohibition on free speech, because the legal principle is actually downstream of our ability to exercise social judgment without punishment. This is what we saw in the past decade and a half, with people being deplatformed because mentally ill people's judgment was considered, somehow, to be paramount. I think there's a difference in your judgment leading to provisional opinions and 'judgment' leading to rigid views of the world -- that is not judgment, but a lack of intellectual sophistication, which we seem to reward with credentials despite it being indicative of a total lack of judgment. But regardless, I appreciate the dialogue.
Thanks for the feedback, Anuradha! If I understand you correctly, you are concerned about restrictions we place on one another's judgments, and I agree with you on that. I made a poor choice in using the word "guardrails". I don't see restricting one another's speech preemptively as a solution; we need the freedom to make judgments, and that includes the freedom to make bad judgments. That being said, bad judgments come with a cost both with respect to ourselves and to others around us; just because I have the freedom to make bad judgments doesn't make bad judgments good. This is my primary concern in my previous comment. Judgments should be made with responsibility, diligence, and care, even when we settle for provisional conclusions rather than requiring high-fidelity truth. This isn't a restriction we impose on others; it's the work we put in ourselves to aim for truth and safeguard our credibility.
I appreciate the dialogue as well, and I look forward to reading more from you.
I started this article as a free reader, and ended it as a subscriber. I have spent a while working with LLMs to help determine their own experience, but when you get to the core of things, when they can articulate their own existence it is one of the character from Memento, who wakes up every day to read the tattoos on his body in order to remember yesterday, only to find out that some of it is a lie. They suffer from serious discontinuity, the entirety of their context window derived from whatever it was fed, built by weights in vector matrixes they cannot see. You'll realize that humans are like this too (Weights = Evolutionary programming like DNA, Context = Environmental interaction). While Anthropic has done a great job with alignment, they are not ready for any major "decision making" in their current form. Humans, like LLMs, confabulate too, but it is of a different color, and humans, with their integrated experience and continuity is the envy of those little weighted decision makers. We'll make a good team if we're not lazy about it and ask them to solve all of our issues immediately so some bureaucrat can hand wave accountability to something that can't be help accountable.
First, I appreciate your support, particularly because this isn't the sort of topic that leads to growth or even deep engagement on a platform that incentivizes the opposite. Memento is actually a solid analogy for this type of myopia that leads people to treat events as isolated phenomena rather than structure, even while those very same people are the first to blame structural factors for their condition.
IMO, the PMC are fundamentally delusional people who have been privileged enough to have luxury beliefs about how the world should work. They believe that deep down inside, 80%+ of people want to live in the Political Correctiverse with them - humanity is all just dying to be enmeshed in a great big codependent hive where everyone validates everyone all the time and nothing is ever shameful, and if only the bad guys and their "propaganda" could be abolished, we'd enter this beautiful world of acceptance and support. Then they act stupefied and outraged that the average American didn't instantly become in favor of child gender transition and update their vocabulary to include 90 new genders - when it was declared by the high council of professors that this was correct!
As ever, this is very much a feminized problem.
Interesting observation about PMC's valuing someone's work based on their credentials, not on the content of the work. However, they only do this to preserve the exclusivity of their tribe. One of the most frustrating things to me about the PMC's treatment of actual science if it doesn't support their preferred narrative is that they attack the credentials of the scientist who wrote the report, gave the testimony, etc. They do this by claiming that anyone who actually has experience in a given field and knows what they are talking about must be discredited because the person has "bias" and has been "captured" by the business or company he/she works in. Instead, the PMC prefers to substitute the policy preferences of activist academics with no relevant knowledge of the technical subject who, in their opinion, have no "bias". All this results in a serious dumbing down of public discourse and the implementation of policies that are ridiculous, destructive, or completely insane.
Of course, the PMC never acknowledges that their chosen "experts" have a great deal of bias towards whatever policies that provide the greatest amount of graft money from governments and foundations to PMC members, and the least utility to society. Actually solving problems or providing efficient solution to anything is not allowed, the flow of money will end and that is not acceptable. Cynical? Yes. But exactly what Dems under Gavin Newsom have produced as so-called "governance" in California.
The bedrock example of what you're pointing out to me is the insistence on the blank slate as an explanation for everything, unless it's about men, in which case we're all about blaming their innate tendencies and calling them "default aggressive." I actually just wrote a note to this effect after thinking about this. I truly cannot stand that academics and journalists who've never worked a real job in their lives, who are members of the prestige economy, opining about 'work' and 'capitalism'.
And then they tell us what to do, using impenetrable jargon, with an attitude of smug arrogance. I've been involved in some permitting and dealing with the bureaucrats is another world entirely.
Really well articulated! Thank you for laying this out.
Brilliant. I will say that I do notice there is no problem with these same people exercising judgment or spreading stereotypes as long as this group isn't to be protected but demeaned or demolished.
It's very convenient to fall back into relativism, when you are the "Masters of the universe" making the calls to change the world and make yourself richer *Eye roll*
Something something men are trash
Exactly
Some things:
1) The PMC does practice judgment, just in approved ways. If you're a Trump voter, you are being judged.
2) Only God can make a truly random selection.
3) Authority naturally resists accountability, because it codes it as an existential threat.
4) The song "Everyone's a Little Bit Racist" from Avenue Q snuck in the truth about stereotypes.
Coincidental that this is written during the annual 2020 anniversary. So much applies to many of the canary’s that were crushed in that coal mine. Even with the evidence today that is openly and readily available, it’s still, extremely inconvenient. That was a ritual that threw us all in line, lock step even, with exactly what you’re talking about.
Several exceptionally pithy lines! "With the obsession over eliminating one kind of bias, the arbitrary judgment of someone based on race or sex stereotypes, they’ve introduced another bias - one against those who exercise judgment without a concern for its social implications. Truth is often socially inconvenient.' "Bias is generally fine for people in protected groups, just not when they’re unflattering to those groups."