Seems your read on ChatGPT is correct. The creators are already putting their thumb on the scale on issues like climate change. The answers to a question like “is climate change more important than grid stability” have shifted in only a few weeks. This issue is complex but if all you get is a tribal answer I’ll pass.
Regarding Twitter, Winegard evidently prefers we all should affect only a complacent equanimity in the face of shocking disclosures of government abuse of power and abject Quisling-like compliance on the part of the Twitter personnel. In the interests of maintaining good relationships with dogmatic ideologues in the hopes they might somehow become amenable to his views, he counsels behavior that is more likely to lead to simply defining deviancy down, i.e., letting matters pass with only some perfunctory polite tut-tutting. When such outrageous violations of basic moral principle and even criminal prohibition have taken place, strong negative consequences are imperative.
"online political discussions attract superficial, status-driven people who act as horridly online as they do offline. "
I suppose your blog attracts people with a relatively narrow range of perspectives and less disagreement (even if I've disagreed with quite a bit) but I like the level of politeness. Merry Christmas.
"But a less pugnacious conservatism could appeal to a broader constituency." If broader means more voters, we could look at a few data points:
Bush 2004: 62 million Romney 2012: 59 m Trump 2016: 62 m Trump 2020: 75 m
I do NOT believe conservatives who fight less, lose more graciously, accept Democrat victories without objection, get more votes. Trump won in 2016 because he FIGHTS. Some Christians hate that; many college educated folk think he's evil for fighting.
That article describes PopUp PACs: >>“Pop-up” super PACs form right before an election so they can spend big now and disclose later, leaving voters in the dark about who is seeking to influence their vote. Worker Power PAC for Georgia’s pre-runoff report covering the period through Nov. 16 does not disclose any donors or expenditures, but instead reports the $1.8 million as a debt and obligation.<<
"Because I place a higher weight on truth than on use of natural language, I think that ChatGPT may be a step backward for AI, not a step forward." Nobody knows the truth about the future - but the true answers to what-if alternative decisions today is what people really want. And if AIs start competing in prediction markets with humans, I believe we can get better predictions.
Don't know if AIs teach Reps to fight Dems more pugnaciously, tho especially more effectively, than they have been so far. Mitch McConnell's midterm actions shows the GOPe is more willing to lose to Dems, than to win with MAGA GOP folks. (It's not just McConnell; Barr and Wray are GOPe too)
Trump's huge increase in votes in 2020, as well as more Black & Hispanic votes, is evidence that fighting has broadened the Republican appeal.
Wanting the "less pugnacious" statement to be true doesn't make it true; and I'm pretty sure it's false. And truth is also important to me. Which is another reason I often read Steve Sailor, who is among the most truthful but also un-PC pundits - whose style and attitude is quite smug and not to the taste of many.
Merry Christmas, I don't want to fight tonight -- it's a Christmas song I sang at Karaoke last time:
My worry about the chat interface is that it distracts from figuring out what these very big models are actually good out, and it discourages developers and product thinkers from building their own experiences.
"I am tempted to stake out a skeptical position regarding GPT-3."
There is an opinion I have seen online. I have seen it from Megan McArdle and numerous professors, and it is that google is not the product or tool that it used to be. It is generally posited that google has not just stagnated, but that it has declined in terms of search functionality.
I wonder whether this is because of the information tsunami or a lack of algorithm advancement, or just a stagnation that people feel should be an improvement due to always being able to imagine better. Either way I live in a world of average people and they are frankly not good at googling anyway and the average person is often borderline innumerate, especially outside their narrow area of expertise. If these language models can work out their accuracy bugs I could see this being a significant increase in utility for average people with regard to internet search. Isn't there some line about Queen Elizabeth already having silk stockings aplenty and the great ability of capitalism was to bring silk stockings aplenty to the masses. Can these natural language models bring a level of language ability to the masses that was previously reserved for the cognitive elite?
This does not make for an AI revolution and your point is 100 percent correct on that, but it will mean a lot of disruption and change regardless of whether google incorporates it or is supplanted eventually.
One possibility - the puckish, the Dionysian - is very helpful and robust at small scales and somewhat ritualized. The Fiddler on the Roof blessing for the Tzar, the local bootlegger whom everybody knows, Halloween, and so on. In short, democracy and populism in less connected, local, deeply knowledgeable [in a way, with tacit knowledge and local insight], traditional... all makes sense. Having a central authority with very limited actual power which is bureaucratic and rulebound is also probably quite useful. Having some tension between them, particularly so that over-reach by the central authority is met with a stern slap locally, but not a viable general revolution, is also probably great. Finally, innovation - locally, innovation can happen very slowly and in a decentralized way, but essentially be stabilizing and so many communities experiment and find solutions for local (sometimes social, often material) problems. Meanwhile, central innovation probably has to happen more rapidly but infrequently, mostly to find solutions to bureaucratic emergent problems.
Seems your read on ChatGPT is correct. The creators are already putting their thumb on the scale on issues like climate change. The answers to a question like “is climate change more important than grid stability” have shifted in only a few weeks. This issue is complex but if all you get is a tribal answer I’ll pass.
Regarding Twitter, Winegard evidently prefers we all should affect only a complacent equanimity in the face of shocking disclosures of government abuse of power and abject Quisling-like compliance on the part of the Twitter personnel. In the interests of maintaining good relationships with dogmatic ideologues in the hopes they might somehow become amenable to his views, he counsels behavior that is more likely to lead to simply defining deviancy down, i.e., letting matters pass with only some perfunctory polite tut-tutting. When such outrageous violations of basic moral principle and even criminal prohibition have taken place, strong negative consequences are imperative.
"online political discussions attract superficial, status-driven people who act as horridly online as they do offline. "
I suppose your blog attracts people with a relatively narrow range of perspectives and less disagreement (even if I've disagreed with quite a bit) but I like the level of politeness. Merry Christmas.
Happy Hanukah, & Merry Christmas to Arnold & all.
"But a less pugnacious conservatism could appeal to a broader constituency." If broader means more voters, we could look at a few data points:
Bush 2004: 62 million Romney 2012: 59 m Trump 2016: 62 m Trump 2020: 75 m
I do NOT believe conservatives who fight less, lose more graciously, accept Democrat victories without objection, get more votes. Trump won in 2016 because he FIGHTS. Some Christians hate that; many college educated folk think he's evil for fighting.
There were many recent races where Trump did not make any recommendation - I see no evidence of more Rep victories there. The GOPe do not fight effectively enough, did not support the fighting Trump supported candidates, many of whom lost with massive Dem party support for the Dem, and minimal Rep Party support (see Walker losing in Georgia to Warnock, $53 mil. vs 150 m https://www.opensecrets.org/news/2022/11/walker-warnock-u-s-senate-race-in-georgia-most-expensive-in-2022-cycle-as-runoff-intensifies/ )
That article describes PopUp PACs: >>“Pop-up” super PACs form right before an election so they can spend big now and disclose later, leaving voters in the dark about who is seeking to influence their vote. Worker Power PAC for Georgia’s pre-runoff report covering the period through Nov. 16 does not disclose any donors or expenditures, but instead reports the $1.8 million as a debt and obligation.<<
"Because I place a higher weight on truth than on use of natural language, I think that ChatGPT may be a step backward for AI, not a step forward." Nobody knows the truth about the future - but the true answers to what-if alternative decisions today is what people really want. And if AIs start competing in prediction markets with humans, I believe we can get better predictions.
Don't know if AIs teach Reps to fight Dems more pugnaciously, tho especially more effectively, than they have been so far. Mitch McConnell's midterm actions shows the GOPe is more willing to lose to Dems, than to win with MAGA GOP folks. (It's not just McConnell; Barr and Wray are GOPe too)
Trump's huge increase in votes in 2020, as well as more Black & Hispanic votes, is evidence that fighting has broadened the Republican appeal.
Wanting the "less pugnacious" statement to be true doesn't make it true; and I'm pretty sure it's false. And truth is also important to me. Which is another reason I often read Steve Sailor, who is among the most truthful but also un-PC pundits - whose style and attitude is quite smug and not to the taste of many.
Merry Christmas, I don't want to fight tonight -- it's a Christmas song I sang at Karaoke last time:
https://www.youtube.com/watch?v=KIhIBFPtnoc
My worry about the chat interface is that it distracts from figuring out what these very big models are actually good out, and it discourages developers and product thinkers from building their own experiences.
"I am tempted to stake out a skeptical position regarding GPT-3."
There is an opinion I have seen online. I have seen it from Megan McArdle and numerous professors, and it is that google is not the product or tool that it used to be. It is generally posited that google has not just stagnated, but that it has declined in terms of search functionality.
I wonder whether this is because of the information tsunami or a lack of algorithm advancement, or just a stagnation that people feel should be an improvement due to always being able to imagine better. Either way I live in a world of average people and they are frankly not good at googling anyway and the average person is often borderline innumerate, especially outside their narrow area of expertise. If these language models can work out their accuracy bugs I could see this being a significant increase in utility for average people with regard to internet search. Isn't there some line about Queen Elizabeth already having silk stockings aplenty and the great ability of capitalism was to bring silk stockings aplenty to the masses. Can these natural language models bring a level of language ability to the masses that was previously reserved for the cognitive elite?
This does not make for an AI revolution and your point is 100 percent correct on that, but it will mean a lot of disruption and change regardless of whether google incorporates it or is supplanted eventually.
One possibility - the puckish, the Dionysian - is very helpful and robust at small scales and somewhat ritualized. The Fiddler on the Roof blessing for the Tzar, the local bootlegger whom everybody knows, Halloween, and so on. In short, democracy and populism in less connected, local, deeply knowledgeable [in a way, with tacit knowledge and local insight], traditional... all makes sense. Having a central authority with very limited actual power which is bureaucratic and rulebound is also probably quite useful. Having some tension between them, particularly so that over-reach by the central authority is met with a stern slap locally, but not a viable general revolution, is also probably great. Finally, innovation - locally, innovation can happen very slowly and in a decentralized way, but essentially be stabilizing and so many communities experiment and find solutions for local (sometimes social, often material) problems. Meanwhile, central innovation probably has to happen more rapidly but infrequently, mostly to find solutions to bureaucratic emergent problems.