Posts

Are rationalists least suited to guard AI? 2020-06-20T17:39:19.140Z · score: 2 (1 votes)

Comments

Comment by lucent on Open & Welcome Thread - July 2020 · 2020-07-05T17:19:50.557Z · score: 4 (3 votes) · LW · GW

Where can I learn more of the frontier thought on schizophrenia? The most compelling theory I've heard is a failure to realize one's internal monologue is one's own, manifested as "hearing voices." However, if I lost that feedback loop and began hearing my thoughts as externally produced, I'd immediately think, "uh oh, schizophrenia" yet this absolutely isn't the case for schizophrenics. What model offers an explanation as to why I'd begin to hear voices yet still maintain I'm not schizophrenic despite that being, to me, an obvious, known symptom?

Comment by lucent on Are rationalists least suited to guard AI? · 2020-06-20T17:39:19.581Z · score: 3 (2 votes) · LW · GW

Could rationalists be the worst people to be put in charge of keeping an AI in the box? We can assume:

  1. Smarter people can convince perfectly rational dumber people of anything, even against their best interests.
  2. Evolution figured out it can't be guaranteed to produce the smartest brain and needs to give everyone built-in safeguards.

I've seen AI box transcripts in which the consensus was: the AI stayed in the box because the AI player was too "dumb" to be swayed by the nuanced, rational arguments. Instead of attributing stubbornness to lack of intelligence, is it a deliberate installation by evolution to defend against clever wordplay and manipulation? Is rationalism the ultimate "cult" as its defining feature is a volitional removal of a safeguard in place to prevent domination of values and goals by those with superior wordplay?

Are there loopholes in this mechanism specifically for one's parents? Is that why the values given to you by your parents, whether about religion or automobile brands, are so difficult to displace? If religion's propagation is primarily through a parental trust loophole closed to others, would that be easy to test by seeing if a preference for Ford or Chevy is just as generationally robust?

Comment by lucent on Open & Welcome Thread - June 2020 · 2020-06-12T05:45:49.632Z · score: 11 (6 votes) · LW · GW

I'm sure this phenomenon has a name by now, but I'm struggling to find it. What is it called when requirements are applied to an excess of applicants solely for the purpose of whittling them down to a manageable number, but doing so either filters no better than chance or actively eliminates the ideal candidate?

For example, a job may require a college degree, but its best workers would be those without one. Or an apartment complex is rude to applicants knowing there are an excess, scaring off good tenants in favor of those desperate. Or someone finds exceptional luck securing online dating "matches" and begins to fill their profile with requirements that put off worthwhile mates.

Comment by lucent on Taking Initial Viral Load Seriously · 2020-04-02T17:31:10.472Z · score: 4 (3 votes) · LW · GW

Is it possible we already have a lot of information about the impact of viral load and just don't realize it? Ever since we switched to agriculture and civilization, the human population has had a large fraction wiped out dozens of times. I don't think we're descendants of merely lucky or biologically "sturdy" people, but of people whose behaviors optimized viral load or had personality traits that helped them avoid it entirely.

Could we be looking at two distinct, competing behaviors which help survival that have evolved out of the ashes of the plagues that devastated humanity? One may be mysophobia/OCD, a behavior pattern about things we cannot see. If this trait has a large hereditary component, that would be strong evidence of why it came about.

The competing strategy may be deliberately attempting to introduce a small viral load frequently to avoid a large load from something like a sneeze to the face. Based on the quantity of recent memes lamenting difficulty in not touching one's face, perhaps there's a genetic component to it. It seems like a stretch to suggest a face-touching gene, but with how many droplet contagious diseases have killed so many of us, one would think it would be selected against by now if it didn't have some redemptive factor.

I'd be most convinced if population modeling showed mysophobia and frequent face-touching were rarely traits that occurred in the same person, suggesting they were opposing strategies, an attempt to avoid exposure entirely and an attempt to deliberately take on a small exposure. Would a combination of both traits compromise survival? Do people at different ages have differing severity of mysophobia or face-touching corresponding to their immune system strength? Does the quantity of each trait in comparison to its population density or type of historical plague correlate strongly enough to these traits to draw any conclusions?

Are we saying "don't touch your face" to people who've specifically evolved to do just that because it got their ancestors through dozens of plagues?

Comment by lucent on March Coronavirus Open Thread · 2020-03-17T15:38:45.709Z · score: 1 (1 votes) · LW · GW

As a child who repeatedly got tonsillitis several times a year, sometimes when totally isolated and sometimes later hearing everyone else around me also got a sore throat, I grew up confused as to why contagious disease tracking was a problem we'd given up on and accepted. No doctor was ever curious of the specific virus or bacteria which caused it. While no one case greatly affected my life, as a whole they very much did. Then again, I was also struck by the possibly false realization that many STDs which have followed us for thousands of years could be permanently wiped out by a couple weeks of global abstinence.

Might we begin to have an intersection of low cost, widely-available tests for every strain of most every contagious agent along with a belief that contagions running amok in civilization is not an unchangeable reality after all this? These at-home DNA tests will soon hit a critical mass of data to link together a global family tree of every human--a previously unthinkable reality that'll suddenly become possible. I'm ready for such a critical mass in contagion tracking that makes a light bulb go off in our collective heads that contagion no longer be an essential part of civilization. Feel slightly sick or have been around someone coughing? We mailed you an envelope of 6 paper swabs for the back of your tongue that'll change color when exposed to the viruses we're aware are circulating in your area.

Learning that a specific RSV strain overwhelmed a daycare and infected 6 babies because my cough I got on a trip to a hotel upstate was cleaned by a chambermaid who got it from... would strongly alter my belief about the inevitability of even minor contagions in the same way the person to person tracking and quarantining of Coronavirus cases is radically altering our social and hygiene landscape right now. Even if widespread testing only shook the "people just get sick for a couple weeks once or twice a year; it happens" attitude of inevitability, it'd be worth it for that alone once it was economically feasible.

Comment by lucent on Anchoring and Adjustment · 2019-11-26T01:39:54.071Z · score: 2 (2 votes) · LW · GW

Is there general agreement that anchoring experiments are a subversion of an evolutionary trait that is generally beneficial? It's rare to be in a group, be presented with a "random" number, and then be asked a question whose answer will be an unrelated number. Unless you have a lot of group status, it's much less harmful to your standing to be wrong with many others frequently than it is beneficial to be right alone infrequently. It's only recent in our evolutionary history that the balance has tipped in the other direction.