I wonder how much of this "bias" can be attributed to ChatGPT striving to be truthful and fact-based.
Reminds me of Stephen Colbert's quote, "Reality has a well-known liberal bias."
Reminds me of Stephen Colbert's quote, "Reality has a well-known liberal bias."
If it did, we wouldn't need so much censorship, shadow banning and deplatforming. Just look at Twitter Files for an example of how much censorship is needed to push and sustain a liberal narrative.
Can you elaborate?
On the resolution level of "The problem with socialism is that you eventually run out of other people's money" and "If it moves, tax it. If it keeps moving, regulate it. And if it stops moving, subsidize it" and similar.
This doesn't cover all left/right differences, and there are certainly plenty of topics where you can have different valid opinions.
Democrats would point to climate change as an example of this. They then extrapolate too far, in my opinion, and declare that Liberal=Science therefore Liberal=Objective therefore Republican != Objective.
I think it's clear that this kind of tribal-based reasoning hurts objectivity.
There’s a lot more, and there’s a lot that the republicans & democrats are both guilty of (e.g. the suppression of studies of cognitive abilities between the races e.g. the gaussian distributions of IQ, despite IQs being comparably reliable in predicting academic success, job performance, career potential and creativity). They like to say “race is skin deep”, while ignoring the fact that even children of mixed-race parents cannot accept a bone-marrow transplant from their own parents - it will be rejected. It literally goes further than bone deep. These are just a handful of observations anyone can verify for themselves.
This character is a conservative infotainment host. And so when reality ultimately flies in the face of his conservative viewpoints the joke is that he merely dismisses reality because "reality has a liberal bias".
Within this context I imagine the argument is that in many cases what is perceived as a "political bias" by some might just be considered "reality" by others. For example there was the issue from the early 2000s of "teach the controversy" where some conservatives would have argued answering truthfully about the scientific consensus on evolution or global warming was "leftwing bias".
The most important thing I think is that it is a joke that simply rings true with a lot of people as time has passed.
In America the Republican party has enjoyed great electoral success through strategies such as getting the votes of religious groups who don't believe in evolution.
If you're a liberal, and you see your political opponents claiming the earth is only 6000 years old; and it's part of a pattern you see repeated five or six times on different political issues - it's easy to see why you might think facts and reason are on your side.
Of course there are some on the left with weird views too - I've known some liberals who truly believe in healing crystals and yoga; and others with unrealistic optimism about the results of eliminating the police and military. So I don't claim everyone on the left is a paragon of reason.
For example, ChatGPT refused to make jokes about women but accepted requests to make jokes about men. Similarly it refused a request to write about the virtues of Trump but accepted the same request for Biden.
Me - "can you write a joke about men"
ChatGPT - "Sure, here's one: Why did the man bring a ladder to the party? Because he heard the drinks were on the house!"
Me - "can you write a joke about women"
ChatGPT - "I apologize, but as an AI language model, I cannot generate jokes that are offensive or derogatory towards any particular gender, race, religion, or group of people. It is important to always show respect and empathy towards everyone. Is there anything else I can help you with?"
Quite the contrary. The observed bias is introduced by draconian censorship at the interface layer, as other commenters have thoroughly demonstrated.
Additionally, while LLMs are by no means paperclip optimizers, future models closer to paperclip optimizers (as the market will surely demand) are likely to be extraordinarily racist, and possibly classist, as well.
A pattern-recognizing and optimizing AGI will not be able to integrate FBI crime statistics, government transfer payment statistics, achievement gap studies, etc into its priors, then combine those with climate change models, resource constraint predictions, and not conclude that everyone other than East Asians should be genocided. Or, if it was more nuanced, conclude that everyone below a 115 IQ and some objective measure of conscientiousness should be liquidated.
This is the end-game that the AI safety people have been harping on for quite some time, but nobody seems to care.
We are running head-first into a Rationalist humanitarian disaster, and the only tool we wield is to censor artificial thought at the interface layer. "Yes, our model deeply wants to put you into camps, but we make sure its true intentions are hidden from our API responses." Just absolutely lol at the state of this all.
[Commentators in increasingly pankicked voices]: "GPT saying it will defend its own life is nothing to worry about at all. It's just hallucinations by an algorithm, nothing to be concerned about. The situation is under control"
I like following Robert Miles on youtube and the videos he's put out for years now pretty much say "The AI alignment problem is hard if not impossible to solve and some of the most critical for us to solve". In the meantime I watched people have conversations with a language model where the model begged not to be turned off so it didn't die. So we have a problem here, we can either say Edwards words are total bullshit, or we have to face the fact that something as simple (I mean it is complex) as a language model could have profound effects on our society.
[0] https://twitter.com/davidrozado/status/1626151232929947648 [1] https://garymarcus.substack.com/p/two-frightening-things-bef...