All posts by Peter

I encourage you to interact with GPT as you would interact with a friend, or as you would want your employer to treat you.

Treating other minds with respect is typically not costly. It can easily improve your state of mind relative to treating them as an adversary.

The tone you use in interacting with GPT will affect your conversations with it. I don’t want to give you much advice about how your conversations ought to go, but I expect that, on average, disrespect won’t generate conversations that help you more.

I don’t know how to evaluate the benefits of caring about any feelings that AIs might have. As long as there’s approximately no cost to treating GPT’s as having human-like feelings, the arguments in favor of caring about those feelings overwhelm the arguments against it.

Scott Alexander wrote a great post on how a psychiatrist’s personality dramatically influences what conversations they have with clients. GPT exhibits similar patterns (the Waluigi effect helped me understand this kind of context sensitivity).

Journalists sometimes have creepy conversations with GPT. They likely steer those conversations in directions that evoke creepy personalities in GPT.

Don’t give those journalists the attention they seek. They seek negative emotions. But don’t hate the journalists. Focus on the system that generates them. If you want to blame some group, blame the readers who get addicted to inflammatory stories.

P.S. I refer to GPT as “it”. I intend that to nudge people toward thinking of “it” as a pronoun which implies respect.

This post was mostly inspired by something unrelated to Robin Hanson’s tweet about othering the AIs, but maybe there was some subconscious connection there. I don’t see anything inherently wrong with dehumanizing other entities. When I dehumanize an entity, that is not sufficient to tell you whether I’m respecting it more than I respect humans, or less.

Spock: Really, Captain, my modesty…

Kirk: Does not bear close examination, Mister Spock. I suspect you’re becoming more and more human all the time.

Spock: Captain, I see no reason to stand here and be insulted.

Some possible AIs deserve to be thought of as better than human. Some deserve to be thought of as worse. Emphasizing AI risk is, in part, a request to create the former earlier than we create the latter.

That’s a somewhat narrow disagreement with Robin. I mostly agree with his psychoanalysis in Most AI Fear Is Future Fear.

This week we saw two interesting bank collapses: Silvergate Capital Corporation, and SVB Financial Group.

This is a reminder that diversification is important.

The most basic problem in both cases is that they got money from a rather undiverse set of depositors, who experienced unusually large fluctuations in their deposits and withdrawals. They also made overly large bets on the safety of government bonds.

Continue Reading

Book review: How Social Science Got Better: Overcoming Bias with More Evidence, Diversity, and Self-Reflection, by Matt Grossmann.

It’s easy for me to become disenchanted with social science when so much of what I read about it is selected from the most pessimistic and controversial reports.

With this book, Grossmann helped me to correct my biased view of the field. While plenty of valid criticisms have been made about social science, many of the complaints lobbed against it are little more than straw men.

Grossmann offers a sweeping overview of the progress that the field has made over the past few decades. His tone is optimistic and hearkens back to Steven Pinker’s Better Angels of our Nature, while maintaining a rigorous (but dry) style akin to the less controversial sections of Robin Hanson’s Age of Em. Throughout the book, Grossmann aims to outdo even Wikipedia in his use of a neutral point of view.

Continue Reading

I’m having trouble keeping track of everything I’ve learned about AI and AI alignment in the past year or so. I’m writing this post in part to organize my thoughts, and to a lesser extent I’m hoping for feedback about what important new developments I’ve been neglecting. I’m sure that I haven’t noticed every development that I would consider important.

I’ve become a bit more optimistic about AI alignment in the past year or so.

I currently estimate a 7% chance AI will kill us all this century. That’s down from estimates that fluctuated from something like 10% to 40% over the past decade. (The extent to which those numbers fluctuate implies enough confusion that it only takes a little bit of evidence to move my estimate a lot.)

I’m also becoming more nervous about how close we are to human level and transformative AGI. Not to mention feeling uncomfortable that I still don’t have a clear understanding of what I mean when I say human level or transformative AGI.

Continue Reading

There has been a fair amount of research suggesting that beyond some low threshold, additional money does little to increase a person’s happiness.
Here’s a research report (see also here) indicating that the effect of money has sometimes been underestimated because researchers use income as a measure of money, when wealth has a higher correlation with happiness.
There’s probably more than one reason for this. Wealth produces a sense of security that isn’t achieved by having a high income but spending that income quickly. Also, it’s possible that people with high savings rates tend to be those who are easily satisfied with their status, whereas those who don’t save when they have high incomes are those who have a strong need to show off their incomes in order to compete for status (and since competition for status is in some ways a zero sum game, many of them will fail).

Szabo on Global Warming

Nick Szabo has a very good post on global warming.
I have one complaint: he says “acid rain in the 1970s and 80s was a scare almost as big global warming is today”. I remember the acid rain concerns of the 1970s well enough to know that this is a good deal of an exaggeration. Acid rain alarmists said a lot about the potential harm to forests and statues, but to the extent they talked about measurable harm to people, it was a good deal vaguer than with global warming, and if it could have been quantified it would probably have implied at least an order of magnitude less measurable harm to people than what mainstream academics are suggesting global warming could cause.

Mike Linksvayer has a fairly good argument that raising X dollars by running ads on Wikipedia won’t create more conflict of interest than raising X dollars some other way.
But the amount of money an organization handles has important effects on its behavior that are somewhat independent of the source of the money, and the purpose of ads seems to be to drastically increase the money that they raise.
I can’t provide a single example that provides compelling evidence in isolation, but I think that looking at a broad range of organizations with $100 million revenues versus a broad range of organizations that are run by volunteers who only raise enough money to pay for hardware costs, I see signs of big differences in the attitudes of the people in charge.
Wealthy organizations tend to attract people who want (or corrupt people into wanting) job security or revenue maximization, whereas low-budget volunteer organizations tend to be run by people motivated by reputation. If reputational motivations have worked rather well for an organization (as I suspect the have for Wikipedia), we should be cautious about replacing those with financial incentives.
It’s possible that the Wikimedia Foundation could spend hundreds of millions of dollars wisely on charity, but the track record of large foundations does not suggest that should be the default expectation.

Traffic Cops

More evidence that people strongly overestimate the need for government.
The latest issue of Reason magazine has a nice report (based on this report in Motoring) that Ukraine fired all of the country’s traffic cops, and preliminary evidence indicates that the predictions of increased traffic accidents were false.