|
Jonathan Mannhart
@
JMannhart
|
|
Cognitive Science @uni_tue. Effective Altruism, Bayes, happiness/mental health, AI Alignment, reward hacking, mountainbiking, reading books, looking for aliens.
|
|
|
456
Tweetovi
|
541
Pratim
|
74
Osobe koje vas prate
|
| Tweetovi |
|
Jonathan Mannhart
@JMannhart
|
21 h |
|
Also, back to your original tweet, I think you have a point. A lot of people seem to report that this value change, oscillating between more/less depressed states, can be experienced as extremely weird. “Wait I should get help this is bad.“ vs. “No, there‘s no point in anything.“
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
21 h |
|
I agree that it‘s interesting. Humans seem to have a weird mix of values and “subvalues“ and they also change over time (which is weird if you think about it from a cs/optimization process perspective). Depression is one of the more unintuitive outputs of this complex algorithm.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
23 h |
|
I can also highly recommend Thrive by Clark and Layard, if you‘re interested in the topic and the effort of pushing those numbers up: amazon.com/Thrive-Evidenc…
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
23 h |
|
|
||
|
Jonathan Mannhart
@JMannhart
|
23 h |
|
Depression isn‘t always “on“, 100%, all the time. It comes and goes in waves (how strong depends on the person and context of course). But getting those numbers higher is one of the most pressing problems in the world today.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
23 h |
|
Hm? I‘m surprised by your prior!
As far as I know most people with depression do not seek help. The 20-40% who do, probably do it because of friends/family getting them to do it, or because one has some better days/hours and getting help is the #1 productive thing to do.
|
||
|
|
||
| Jonathan Mannhart proslijedio/la je tweet | ||
|
Robert Wiblin
@robertwiblin
|
4. velj |
|
"Don't worry about Coronavirus more people died of flu today."
"Don't worry about the flu, the number of people infected isn't rising 50% a day and we know its case fatality rate is low."
I know which I find more persuasive.
|
||
|
|
||
| Jonathan Mannhart proslijedio/la je tweet | ||
|
Kieran Healy
@kjhealy
|
3. velj |
|
Never use Academia-not-edu or ResearchGate. If you want to share your working papers, put them on your website or use an open archive. It's a disgrace that the academia thing even has a .edu domain. twitter.com/johnlaudun/sta…
|
||
|
|
||
| Jonathan Mannhart proslijedio/la je tweet | ||
|
Robert Wiblin
@robertwiblin
|
30. sij |
|
.@OurWorldInData has really upped their game lately. Their pages are so damn useful, e.g. see below.
Congrats @MaxCRoser!
ourworldindata.org/co2-and-other-…
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
30. sij |
|
Amazing!!
|
||
|
|
||
| Jonathan Mannhart proslijedio/la je tweet | ||
|
Haydn Belfield
@HaydnBelfield
|
23. sij |
|
#DoomsdayClock is closest its ever been to midnight
People asked me "can that really be true?". I'm a Cambridge Uni existential risk researcher and my view is: unfortunately yes
TL:DR nuclear situation worse than last 30 yrs, climate situation worse ever, and new bio + AI risks
|
||
|
|
||
| Jonathan Mannhart proslijedio/la je tweet | ||
|
Faiz Memon
@_faizmemon_
|
22. sij |
|
This was one of the most depressing and yet motivating data I've seen on this topic of who we spend our time with and it's finiteness.. pic.twitter.com/WkG1FIib73
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
19. sij |
|
Exactly. I think that‘s a logical fallacy. The question is not if learning X is *any* good at all, the question is the counterfactual: Is it better to learn X than to learn Y. We could say “but people should learn/know more“, but in reality we need to prioritise, and do it well.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
16. sij |
|
Oh boy! Would love to see that!
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
16. sij |
|
What will be important waypoints/milestones for AI development in the future in regards to risk? Which inventions/events should influence our risk assessment and in which direction and how much?
And what are his predictions regarding these things happening sooner or later?
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
16. sij |
|
Does he think cognitive psychology/moral psychology is relevant to the alignment problem? I heard Eliezer Yudkowsky saying that >99% of the work is getting the cs/math part right and not the “figuring out human values“ part. His opinion on this could influence my career choices.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
12. sij |
|
This was recommended to me by some predictive processing folks in Sussex, haven‘t come around to reading it, but I hold their recommendations in very high regard. Also, Dan Dennett likes it: amazon.com/Dreaming-Conce…
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
10. sij |
|
After that it‘s really quite easy to go for a 30 or 60 minute run, because you actually want to do it. This takes some time, so — super important — don‘t be angry at yourself if you just don‘t like it for some time. But getting to the point when it‘s fun, that’s the actual goal.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
10. sij |
|
I still do this, I think it‘s the biggest help with motivation. Set some time (30 minutes) and it‘s *totally okay if you walk most of the time*. The habit counts. After some time, running will become easier, and if you run for 20 of 30 minutes AND have fun, you‘ve won already.
|
||
|
|
||
|
Jonathan Mannhart
@JMannhart
|
10. sij |
|
And at last the most important thing in my opinion: just start running, but walk if you don‘t want to run anymore. There is nothing worse for your brain to train it that running hurts too much (at first, later one can learn to enjoy the exhaustion). But just walk for a while.
|
||
|
|
||