David Manheim @davidmanheim Twitter profile

Last Seen Profiles

@KDSportsReport

@hannah_mclaugh

@therealahype1

@zilke_YouTube

@gabrielisgayasf

@hbouammar

@BrianEtema35362

@TAEYEONTHAI

@7bp__

@TMiyaso

@TScooby2

@zevza0

@aparnapkin

@ganno_satoshi

@MiL_cos

@donttakeitpls

@sara_dillon

@JNFRNCH

@virginmobilekw

@MelisaQuintan20

@dymondtarane

@NcaaFBall25

@parkbh878

@janfrodeno

@V5b5gB

@finetipelmig

@kieraneason

@ExoticPapers

@RakshanVJ

@HoweAmy34664

@asioigle2013

@JohnStossel

@CAPEOO7

@dyfttt01

@sakeyake78

@Codie_fortunate

David Manheim

@davidmanheim

13 days

@luxxnoirlondon Current ick: People feeling the need to have opinions about, much less be bothered by, other people's behaviors that affect them in no way at all.

luxx noir london ⋆｡°✩

@luxxnoirlondon

14 days

Current ick: men presumably under 35 reading books on the subway or in public.

816

43

896

11

65

3K

David Manheim

@davidmanheim

4 years

I feel personally attacked.

3

160

870

David Manheim

@davidmanheim

3 years

Humanity spent millennia haunted by waterborne illnesses like cholera, typhoid, and dysentery. Then we figured out how to make water safe. We are still plagued by airborne diseases like COVID-19, influenza, and common colds. Giving everyone access to safe air is an obvious goal.

58

86

549

David Manheim

@davidmanheim

3 years

The year is 3627. All stars within 1,000 light years of Earth are enclosed in Dyson spheres, and all sentience has merged into an over-mind. Printer drivers are over 70 exabytes, and can take weeks to set up. The over-mind wonders if this can ever be fixed.

18

67

481

David Manheim

@davidmanheim

3 years

@nicholaraihani @lakens We need to normalize naming and shaming journals and editors for exacerbating the replication crisis and file-drawer biases for published research.

5

12

454

David Manheim

@davidmanheim

2 years

"How can we create an AGI if we don't even know what intelligence even is?" is trying to focus on having a mysterious question, assuming that it needs a mysterious answer. In fact...

11

28

423

David Manheim

@davidmanheim

3 years

It's not that I insist on using 24 character passwords - it's just the default I use for generating passwords. But if your site says the password is too long, I immediately assume that the rest of your security infrastructure is also a joke.

6

41

356

David Manheim

@davidmanheim

2 years

The recent (inexplicable and irresponsible) decision by the NIH to fund more of Peter Daszak's dangerous bat-virus discovery work brings up the obvious, and easily answered, question; is this safe? No, it's not, in two different ways. Thread, 1/6

William Eden

@WilliamAEden

2 years

Peter Daszak has received another grant from the NIH… …to study bat coronaviruses in the wild. After everything the world has just been though. After all the risky research that was supposed to protect us from a global pandemic failed to stop one.

86

429

2K

23

76

340

David Manheim

@davidmanheim

2 years

People interested in reducing biorisk seem to be super excited about 222nm light to kill pathogens. I’m also really excited - but it’s (unfortunately) probably a decade or more away from widespread usage. Let me explain.

11

38

328

David Manheim

@davidmanheim

1 year

@AlecStapp Correlation is strong evidence of causation, but (without other data, like an understood mechanism) it isn't evidence of where the causality is - it might indicate a shared cause, or be a more complex causal relationship.

7

15

279

David Manheim

@davidmanheim

2 years

"Haha, yes, we do know that when we optimize for engagement, we're incidentally optimizing for group rage." It's a fantastic, if disappointing, example of how optimization goes poorly.

Elon Musk

@elonmusk

2 years

Trashing accounts that you hate will cause our algorithm to show you more of those accounts, as it is keying off of your interactions. Basically saying if you love trashing *that* account, then you will probably also love trashing *this* account. Not actually wrong lol.

20K

17K

186K

9

41

249

David Manheim

@davidmanheim

4 years

Israel paid $24/dose for vaccines, $48/person. 3x what the US paid. It was a bargain. International vaccine equity is important, and Israeli lives certainly aren't worth more than lives anywhere else in the world - but their government was willing to pay more to protect them.

15

36

245

David Manheim

@davidmanheim

2 months

Hats off to Isaac Asimov for correctly predicting exactly this 75 years ago in I, Robot: Some people won't accept anything that doesn't eat, drink, and eventually die as being sentient.

Ana Brandusescu

@anabrandusescu

2 months

Because it bears repeating: "An LLM is a mathematical model coded on silicon chips. It is not an embodied being like humans. It does not have a “life” that needs to eat, drink, reproduce, experience emotion, get sick, and eventually die."

20

25

92

14

18

196

David Manheim

@davidmanheim

7 months

@calebwatney @peterwildeford Still massively underpaying people. The GS levels are not enough for skilled talent in AI.

8

3

228

David Manheim

@davidmanheim

3 years

Charity pledges: Elon Musk: Giving Pledge (GP), 50% Jeff Bezos: No pledge (None) Bernard Arnault: None Bill Gates: GP, ~100% Mark Zuckerberg: Non-specific, 99% Larry Page: None Sergey Brin: None Steve Ballmer: None Larry Ellison: GP, 95% Warren Buffett: GP, 99%

Jon Erlichman

@JonErlichman

3 years

Net worth Elon Musk: $205 billion Jeff Bezos: $200 billion Bernard Arnault: $159 billion Bill Gates: $151 billion Mark Zuckerberg: $136 billion Larry Page: $126 billion Sergey Brin: $121 billion Steve Ballmer: $107 billion Larry Ellison: $101 billion Warren Buffett: $101 billion

140

324

2K

7

29

221

David Manheim

@davidmanheim

4 years

A Simple Proposal for Jumpstarting Vaccine Production In about 12 months, the world will need to start producing massive quantities of a COVID-19 vaccine, but we don't have enough vaccine production facilities to quickly produce the billions of doses we need globally. (1/9)

8

42

214

David Manheim

@davidmanheim

3 years

#QOTD : It's very popular to say things like "It's wrong to profit off of human suffering." Somehow, people never mean that it's wrong to profit from *causing* human suffering when they say this; they mean that it's wrong to profit from *preventing* human suffering - @AlexMennen

8

30

217

David Manheim

@davidmanheim

1 year

OK, but no-one would be stupid enough to deploy this publicly, with unrestricted access to the internet... would they? They did? Well, I'm hoping they learn their lesson and won't do it with more powerful systems. They almost certainly will? Humanity is so doomed.

6

20

210

David Manheim

@davidmanheim

2 years

Thankfully, the accelerated timeline that Trump pressured them for meant vaccines were available within a year, and far fewer people died than would have if no-one were pressuring the FDA. Trump did many, many things wrong - don't attack him for what he did right.

POLITICO

@politico

2 years

The Trump administration pressured the FDA to authorize unproven treatments for Covid-19 and the first Covid-19 vaccines on an accelerated timeline, according to a House report released Wednesday.

5K

768

2K

7

22

188

David Manheim

@davidmanheim

4 years

@masonporter @KellerScholl I'm definitely citing this. "The literature on COVID-19 is rapidly growing (Lit review 1, 2, 3) but only some of it is useful or even meaningful. (Porter, 2020)"

Mason Porter

@masonporter

4 years

I am here to help.

94

3K

10K

0

9

183

David Manheim

@davidmanheim

1 year

To predict it in advance: By 2025, people will point to the fact that AI has continued to accelerate, but that most of the harms were of the types being predicted about employment and bias, and gleefully proclaim that @ESYudkowsky was wrong about the real risks.

24

9

185

David Manheim

@davidmanheim

2 years

Does anyone know of a case in the last century where the argument against that not everyone will get access to a new technology (rather than via accruing capital / wage disparity,) wasn't falsified by falling prices?

Pessimists Archive

@PessimistsArc

2 years

CNN's first report on the internet in 1994 aired concerns it was 'elitist': “Critics say one danger is this could turn out to be an elitist system, one available only available to people with a computer and a modem. And thus widen the gap between the haves and the have nots.”

16

56

237

24

19

182

David Manheim

@davidmanheim

1 year

Is @ylecun suggesting we need licensing of every ML developer, with a federal agency tracking every single AGI run, with investigation into every accident and near-miss, in order to build confidence in safety of AGI? (Because if so, I think he's really on to something here!)

Yann LeCun

@ylecun

1 year

Some folks say "I'm scared of AGI" Are they scared of flying? No! Not because airplanes can't crash. But because engineers have made airliners very safe. Why would AI be any different? Why should AI engineers be more scared of AI than aircraft engineers were scared of flying?

1K

431

4K

10

20

182

David Manheim

@davidmanheim

3 months

Primary endosymbiosis found! Tremendously important discovery which significantly updates estimates of the likelihood of evolution of eukaryotic cells, with implications for the great filter hypothesis and alien life. cc: @robinhanson @anderssandberg

Scientists Discover First Nitrogen-Fixing Organelle

After years of work, an international team found evidence that a once-independent nitrogen-fixing microbe has become a permanent resident within algae cells

newscenter.lbl.gov

8

30

180

David Manheim

@davidmanheim

5 years

It is not from the malevolence of the financier, the politician, or the bureaucrat that we expect our systemic failures, but from their regard to their own interest.

6

32

169

David Manheim

@davidmanheim

4 years

Someone should start an annual $1m prize for whoever violated government procedures to do the most good for the world. cc: @paulg / @albrgr / @ESYudkowsky / @robinhanson / @bryan_caplan

15

19

168

David Manheim

@davidmanheim

6 years

Metrics can be used well - they often aren't. And many of the issues created by their misuse can be addressed by better design. In case you haven't seen me mention it yet, here's my new paper on doing metric design more thoughtfully:

9

44

162

David Manheim

@davidmanheim

3 years

Interesting. I would not have predicted that.

Ellen Peters

@ellenpetersjdm

3 years

You can see it here. Life satisfaction doesn't change much past about $50k for the less numerate. But keeps rising (indefinitely?) for the highly numerate. @jasonzweigwsj

22

21

209

17

14

159

David Manheim

@davidmanheim

5 months

Whether or not it truly exists, there is math that describes how it behaves.

Emmett Shear

@eshear

5 months

If it truly exists, there is math that describes how it behaves

119

70

797

12

6

158

David Manheim

@davidmanheim

4 years

Look, AI isn't magic, it's not just going to fix all of your problems. You can't just add AI to an arbitrary systems and make it do amazing new things. You're thinking of *blockchain*

8

14

154

David Manheim

@davidmanheim

3 years

#QOTD : "I really wish the typical person on the left better understood the positive externalities of free markets. I really wish the typical person on the right better understood the negative externalities of free markets."

7

18

149

David Manheim

@davidmanheim

4 years

Well, @slatestarcodex potentially getting doxxed by the @nytimes and taking down the site wasn't something I even had on my list of "what could happen when 2020 gets really crazy."

2

8

144

David Manheim

@davidmanheim

8 months

Regulators are the same as any other group in the world - 0%. Because no-one actually understands the way deep neural networks work. Yes, some people know how to train them, others know how to do prompt engineering and fine tuning, but no-one understands the models themselves.

Austen Allred

@Austen

8 months

What % of the people in this photo do you think understand how AI actually works?

350

159

3K

7

16

121

David Manheim

@davidmanheim

7 years

@juliagalef Cobra Effect FTW!

3

6

138

David Manheim

@davidmanheim

6 years

Israeli elections were just called for April. Here's an amusing and fairly accurate guide to what the Israeli political parties are all about, via @reddit . #IL #IsraElex19

8

46

134

David Manheim

@davidmanheim

3 years

Happy Petrov Day yet again, with wishes to all that humanity coordinates and cooperates with each other enough to survive to celebrate long into the far future.

David Manheim

@davidmanheim

8 years

If you're not already aware, take a minute to Google Stanislav Petrov. h/t @ESYudkowsky Happy Petrov day!

2

7

22

1

32

136

David Manheim

@davidmanheim

6 years

We don't see time travelers. This means either time travel is impossible, or humanity doesn't survive. Evidence of the theoretical plausibility of time travel is therefore strong evidence that we will be extinct in the nearer term future. (cc: @anderssandberg )

37

41

136

David Manheim

@davidmanheim

1 year

One problem existing doesn't mean another problem isn't also real. Working on biorisk in 2017, I heard exactly the same argument: "Focusing on hypothetical pandemics that could kill millions takes attention away from real-world diseases." True, but very bad planning!

Sasha Costanza-Chock

@schock

1 year

Also, catastrophism and concerns over potential 'existential risk' primarily serve to draw attention away from existing real-world harms from AI systems, harms that already occur everyday & fall disproportionately on those who are marginalized within the matrix of domination.

7

27

133

1

13

136

David Manheim

@davidmanheim

4 years

To put this in post-election-pundit terms, there's a 87% chance that "despite the overly cautious so-called predictors, a Biden victory was 100% inevitable," and a 13% chance that "it's an upset that @FiveThirtyEight all-but-guaranteed wouldn't happen."

2

29

136

David Manheim

@davidmanheim

1 year

People confuse nuclear explosions with true chemical explosions, thinking that they might replace chemical explosives. But nuclear weapons can't have general explosive power. Instead, they only emulate impacts of chemical explosions using a fundamentally different process.

9

7

129

David Manheim

@davidmanheim

3 years

Are there lab accidents and other anthropogenic disease sources? Yes. Are some of those events of a severity that could lead to outbreaks? Also yes. Here's our newly published list of 71 such incidents from 1975-2016 in @F1000Research :

2

39

128

David Manheim

@davidmanheim

3 years

How well calibrated were global catastrophic risk experts on the size of future natural pandemics? From Bostrom and @anderssandberg informal 2008 survey, the experts estimated a mean probability of 60% that a natural pandemic would kill 1m people before 2100.

9

17

124

David Manheim

@davidmanheim

5 years

@JimDMiller You could be really meta, and model this problem itself as a game theoretic problem as an example in your class...

James Miller

@JimDMiller

5 years

My game theory class uses examples from dating. If I don't include examples with lesbians, students complain in course evaluations. But students seem uncomfortable when I (middle age straight man) discuss lesbian dating with them. Almost all my students females between 18-21.

100

57

762

1

2

120

David Manheim

@davidmanheim

7 years

Fictional inspiration for scientific speculation moving from Heinlein to Lovecraft is worrisome. cc: @anderssandberg

9

62

119

David Manheim

@davidmanheim

2 months

Scott: I spent 20 years telling people: No, sorry, quantum computing probably can't help your application, and if you give me time, I can explain why. Now, I tell people: Generative AI probably will greatly change your application in the near future, and no, I can't explain why.

David Manheim

@davidmanheim

2 months

Very excited to hear Scott Aaronson talk about what he's been doing in AI Safety, and interesting open problems and research approaches for the students and faculty at @TelAvivUni today.

3

0

36

5

10

126

David Manheim

@davidmanheim

2 years

AI-Risk Skeptics: "Poorly-aligned AI can't pose a real threat to humanity because it's digital, and its impacts are limited to digital ones." @DeepMind : "Hey, look everyone, now we have an ML system directly running the containment system for a nuclear fusion reactor."

11

7

125

David Manheim

@davidmanheim

2 years

@MatjazLeonardis I love the story about Gates foundation (IIRC) making promises to developing countries that they could get AIDS medication at low prices if they committed to buy it, and making promises to pharma that they could get high volumes if the price came down - then not spending a cent.

2

7

120

David Manheim

@davidmanheim

10 months

I don't understand nuclear physics. I don't want nuclear wars. I don't understand the chemistry of methane clathrates. I don't want runaway climate change. I do happen to understand lots about AI - but that's not the point. I don't want dangerous AI systems built.

sarah

@littIeramblings

10 months

And the simple matter of the fact is I don’t understand it. I’m not smart enough, it’s beyond me. Any argument about what an AI system may or may not do, or how afraid of it I shouldn’t or shouldn’t be, sounds just as compelling to my untrained ear as any other

5

1

134

9

5

120

David Manheim

@davidmanheim

3 years

Scientists: Science requires transparency, openness, and verifiability! Also Scientists: Just submitted my research to a pay-for access journal. I used proprietary software, don't know which version. Data and code available on request if it isn't deleted once I get a new laptop.

1

11

117

David Manheim

@davidmanheim

1 year

@Helenreflects It's not the platform - most of the ones who are jerks do it in person too.

3

1

113

David Manheim

@davidmanheim

10 months

Personal update: Me and my family are all OK. And without getting into the politics, our hearts go out to the Palestinians and Israelis and the families and friends of those who have been hurt and killed.

3

0

116

David Manheim

@davidmanheim

1 year

OK, now it's pretty clear to me that this #ChatGPT exploit is getting responses to other people's prompts:

29

9

113

David Manheim

@davidmanheim

3 years

Maybe I sound like a broken record, but it seems the developed world STILL, somehow, never tried throwing lots of money at the parts of the system that constrain supply. Nor purchasing vaccines for the rest of the world to stop variants. (It's a great deal, given the costs.)

Seth Berkley

@DrSethBerkley

3 years

To tackle #COVID19 vaccine supply bottlenecks we need to do everything in our power to increase manufacturing capacity. As @Gavi Board Chair @JMDBarroso outlines: "it’s not enough to share intellectual property, you also need knowledge transfer”:

2

16

45

3

20

110

David Manheim

@davidmanheim

1 year

Is AI a risk? @pmarca says no. What would change his mind? He says that after a rogue AI develops and deploys a custom unique pathogen that has started killing people, he'd reconsider. Only a mass murder of unbounded size could convince him. That's a ludicrously high bar.

16

3

105

David Manheim

@davidmanheim

4 years

Hey, @TwitterSupport / @Twitter - it looks like someone suspended @lukeprog . He works at @open_phil , and tweets about history, AI policy, music, and similarly non-problematic things. This is a silly mistake. @georgetownsfs , @openAI , @FHIOxford - anyone know who can fix this?

7

4

106

David Manheim

@davidmanheim

7 years

@ron_miller @juliagalef @alex Not a myth, rather an inaccurate phrasing. He was paid by the installment, but those installments had length requirements. So he *was* paid for verbosity in a way that produces the same incentives, albeit not literally per word. cc: @Stephen_Sawchuk

5

12

100

David Manheim

@davidmanheim

2 years

It was insanely irresponsible to do this type of gain-of-function work, but publishing it is beyond irresponsible.

William Eden

@WilliamAEden

2 years

This lab from Boston University is taking the Omicron spike protein and adding it to wildtype COVID: “In K18-hACE2 mice, while Omicron causes mild, non-fatal infection, the Omicron S-carrying virus inflicts severe disease with a mortality rate of 80%.”

179

282

1K

8

16

102

David Manheim

@davidmanheim

2 years

A decade ago, surveys about AI timelines by Bostrom and Sandberg, and by Baum, suggested a median of 2050 for an estimated 50% chance of general human-level performance. Are there still experts who believe that high-level machine intelligence is less than 50% likely by 2050?

16

5

100

David Manheim

@davidmanheim

5 years

@brettb13 @qkate @mmmtravis Let's say you're looking for Waldo, and someone pours paint on the page. Just because you know the paint is blue doesn't mean you can see through it and find Waldo.

1

2

94

David Manheim

@davidmanheim

10 months

Happy Petrov Day! As always, wishes to all that humanity improves on our spotty track record of coordination and cooperation with each other enough to survive to celebrate the future.

David Manheim

@davidmanheim

8 years

If you're not already aware, take a minute to Google Stanislav Petrov. h/t @ESYudkowsky Happy Petrov day!

2

7

22

2

17

98

David Manheim

@davidmanheim

4 years

Prediction: By the end of the year, we'll see economic papers and estimates that consider the economic impact of lockdowns which account for deaths and hospital costs, but assume that economic activity without lockdows would have stayed at the previous level.

6

12

95

David Manheim

@davidmanheim

3 years

1) We never had evidence that vaccines don't prevent transmission. 2) We now have evidence that they do help prevent transmission. 3) The science community's "caution" about conclusions plus media's scaremongering with "we don't know if..." is going to kill people. Again.

2

24

97

David Manheim

@davidmanheim

4 years

1

14

97

David Manheim

@davidmanheim

1 year

People argue: "China and Russia won't stop their AI research, so the west needs to allow unsafe work just to compete." This seems like a confused model of the world. Russia doesn't have an AI industry to speak of, and as @ezraklein recently pointed out, China is being cautious.

11

14

96

David Manheim

@davidmanheim

3 years

After reading a bit of what philosophers say about infinite ethics, I'm tempted to write "Why Philosophers Should Care About Real Analysis." SECTION 1: LIMITS ARE A THING SECTION 2: NO, INFINITY ISN'T A NUMBER SECTION 3: HOW CONTINUITY (DOES NOT) WORK

13

8

96

David Manheim

@davidmanheim

3 years

Excited to announce that I'll be a visiting professor this year at @TechnionLive , working to build interdisciplinary collaborations with technical researchers on the promises and risks of emerging technologies.

14

2

94

David Manheim

@davidmanheim

1 year

Remember when the term AGI was used in contrast to ANI, to indicate ability to generalize and show capabilities in areas a ML system wasn't explicitly trained on, instead of meaning "at least human-level in every single imaginable task"? Pepperidge Farm remembers.

9

5

95

David Manheim

@davidmanheim

29 days

Prediction markets don't create causal estimates! An example? The market probability that Democrats win if Newsome replaces Biden combines counterfactual worlds where Biden gracefully steps down, where Dem leaders strongly back Newsome, or where Biden has a heart attack.

8

2

95

David Manheim

@davidmanheim

4 years

The medical and scientific community is concerned about damaging trust in vaccines and in science by taking risks responding to COVID-19. Instead, they damaged trust by failing to do so, avoiding risk and being slow and conservative to react. (Thread.)

2

22

94

David Manheim

@davidmanheim

8 months

@GaryMarcus "GPT-5 isn’t here after a year despite immense commercial desire." There was a 3 year gap between GPT-3 and GPT-4, and they said within the last 6 months that they haven't begun training it, so it isn't saying very much.

2

1

94

David Manheim

@davidmanheim

4 years

Even if you consider the risk of human extinction to be finite, but low, extinction is inevitable over time. A bit of reflection about convergence makes it clear that for humanity to survive over the long run, this probability needs to be both low *and* dropping over time.

10

92

David Manheim

@davidmanheim

4 years

@drklausner @mugecevik @joshmich @DocJeffD Most of the cases being looked at were under conditions that forbid outdoor contact, so it's harder to say what the counterfactual case would be if things were open. That's not to say governments shouldn't open beaches, but it's far less clear than your tweet implies.

0

2

90

David Manheim

@davidmanheim

6 years

"By distributing a small number of good university places based on test scores we could, if we wanted, motivate kids to ruin their entire childhoods feverishly memorising useless information." - @robertwiblin

7

19

90

David Manheim

@davidmanheim

7 years

"Cryptography is nightmare magic math that cares what kind of pen you use." Here's a great example of @SwiftOnSecurity 's point:

1

54

89

David Manheim

@davidmanheim

3 years

A proposal: Let's start preparing for the next pandemic by eliminating diseases now. (Thread)

5

28

89

David Manheim

@davidmanheim

3 years

Realizing that many of the people most worried about existential risks during the short term future are among the most optimistic about humanity and the potential for the long term future.

15

6

86

David Manheim

@davidmanheim

1 year

What people would have called AGI in 2016 - systems that can perform most intellectual tasks better than the average human - is significantly less capable than GPT-4.

François Chollet

@fchollet

1 year

Or closer to the present -- like how people in 2016 predicted that RL applied to game environments would lead to AGI within 5-10 years

9

20

365

9

2

87

David Manheim

@davidmanheim

1 year

@FellowHominid You might want to come up with better ways of phrasing things. "A professional acquaintance that I've been collaborating with in conversations about epistemics. I met them via shared interests on different forums and we've spoken, but have not yet had a chance to meet in person."

1

88

David Manheim

@davidmanheim

10 months

I've expressed support for the Palestinian people in the past, and will continue to insist that Israeli policy is often unacceptable. But if the past couple days led you to loudly express support for the Palestinians, you're not supporting them, you're cheering for terrorism.

2

90

David Manheim

@davidmanheim

3 years

Amazing to see that, given the cost estimates, I've saved multiple lives since I started donating to @GiveWell . All of you can do the same - there are solvable problems in the world. Funding these interventions will save lives effectively: via @TomChivers

Lee Crawfurd

@leecrawfurd

3 years

This is an astonishing set of 2020 results for @GiveWell - Saved 24,000+ lives - Gave 6m+ kids anti-malaria meds - Gave Vit A to 8.6m+ kids - Delivered 4.4m+ mosquito nets - Vaccinated 118,000 kids - Treated 11.4m+ kids for worms

3

30

171

1

16

89

David Manheim

@davidmanheim

3 years

Anyways, I'm sure there is no reason to think Omicron is a lab-escape variant.

9

10

86

David Manheim

@davidmanheim

2 months

Question for @sama : Does @OpenAI have non-disparagement agreement with board members or former board members? If so, is @sama willing to publicly release the text of any such agreements?

4

9

81

David Manheim

@davidmanheim

1 year

Of the 10 non-deleted responses - which narrows the sample considerably - I would judge that 7 didn't happen, and 3 did. So if you had made every bet "offered" in response to this thread, you'd be ahead quite a lot of money.

Eliezer Yudkowsky ⏹️

@ESYudkowsky

7 years

AI hype busters: What would you bet at 9-1 cannot *possibly* be done in 1, 2, or 5 years? (Concrete verifiable tasks only, obviously.)

12

8

19

3

5

87

David Manheim

@davidmanheim

4 years

I have updated greatly towards @bryan_caplan 's views about grade school education, and have found that my kids are happier and learn more at home in far less "learning" time. Also, I now disagree with him completely, because how on earth does he manage to get any work done?

5

7

86

David Manheim

@davidmanheim

4 years

I'm not saying 2020 is turning around, but I certainly didn't anticipate "Pope prays for AI safety" for November.

1

4

87

David Manheim

@davidmanheim

4 months

OK, I figured out the perfect solution: When I have both a footnote and an exponent, I just multiply them to save space!

David Manheim

@davidmanheim

4 months

The difference between footnotes and exponents is really pretty important, and I don't think there's a clear typographical convention for making this clear...

8

0

43

5

9

84

David Manheim

@davidmanheim

2 years

I've been thinking about how people change the world for the better for quite a while. Turns out it's hard, and the world is complex, but more critically, most people aren't trying. And if they care about the world, and want it to be better, that's a shame. (1/25)

3

18

81

David Manheim

@davidmanheim

3 years

Sadly, it seems @Sci_Hub is no longer getting new papers uploaded - but it still has older papers. ( @ringo_ring - ?) Not sure how long this will last, but someone should preregister an instrumental variable paper on scihub availability's impact on citations. #Econometrics

4

15

85

David Manheim

@davidmanheim

6 years

@WeedenKim @bechhof Ditto for "required" job qualifications, and the word no as it applies to salary negotiations. Our unquestioned social defaults have real impacts, and without active compensation, they perpetuate disadvantages.

Kim Weeden

@WeedenKim

6 years

What an administrator says: "this is the rule, but students can petition for an exemption." What a sociologist hears: "this is the rule, but middle- or upper-class students who grew up thinking it's their right to question school authority figures can petition for an exemption.

52

2K

6K

0

9

79

David Manheim

@davidmanheim

3 years

Basic economics should be required for anyone getting a high school diploma, much less a BA. Failing to understand basics of how your economy works is like failing to understand the basics of your political system - it makes you unable to function as a well informed citizen.

8

9

82

David Manheim

@davidmanheim

4 years

@paulg @stewartbrand This is correct. But if large, fragile, and poorly managed companies were allowed to fail instead of being bailed out, we could have a market that rewarded planning for potential disruption - and the economy would likely be much more robust.

Paul Graham

@paulg

4 years

@stewartbrand It is an interesting point that Adam Smith's invisible hand doesn't design for system-wide robustness. Curiously, this means that the US economy might have been more resistant to shocks back in the mid 20th century, when it was composed of a small number of oligopolies.

6

2

47

3

15

84

David Manheim

@davidmanheim

3 years

Reminder: there's a huge gap between experts often being wrong and the idea that someone else's opinion is more likely to be correct.

3

13

84

David Manheim

@davidmanheim

1 year

Wait, why is this only a concern for open-source LLMs? Jan must be pretty damn confident in the cybersecurity and ability to prevent misuse by all the non-open-source labs and university research groups, if he's worried about open-sourcing models, not building them initially.

Jan Leike

@janleike

1 year

An important test for humanity will be whether we can collectively decide not to open source LLMs that can reliably survive and spread on their own. Once spreading, LLMs will get up to all kinds of crime, it'll be hard to catch all copies, and we'll fight over who's responsible

186

55

410

4

8

81

David Manheim

@davidmanheim

2 years

@ESYudkowsky You're only noticing the things that inconvenience you - which is a very small portion of what government does. Most of government is making sure there are roads, schools, scientific research, a military, and courts, running social security and medicare, etc.

7

0

81

David Manheim

@davidmanheim

3 years

The age structure of COVID-19 mortality / severity seems to imply that moderately serious novel infectious disease may be more historically common than previously assumed. (Short thread, including some speculation.)

1

7

79

David Manheim

@davidmanheim

2 years

"How can linear algebra and max functions create intelligence? It's just doing lots and lots of mathematics?" is the same claim as "How can physics create intelligence? It's just doing lots and lots of physics?" If humans are intelligent, (which is arguable,) then ML can be.

6

4

80

David Manheim

@davidmanheim

2 years

A flashback to mid-March 2020. I was dead wrong, following the consensus, and Joe Rogan and his guest were right.

Michael Tracey

@mtracey

4 years

The infectious disease expert on Joe Rogan says hand-washing isn't even effective because coronavirus is transmitted through the air; advising people to wash their hands is just to make them feel like they're "doing something." Great

194

171

1K

1

4

80

David Manheim

@davidmanheim

1 year

@Rainmaker1973 @paulg ...and at the displayed resolution of 768x768, they are racing towards each other at the incredible rate of around 7 million years/pixel!

Massimo

@Rainmaker1973

1 year

- Estimated diameter of the Milky Way: 100,000 ly - Estimated diameter of M31: 220,000 ly - Estimated distance between the two: 2.5M ly This is a realistic representation of the size and distance between the Andromeda Galaxy and Milky Way [source: ]

104

471

4K

0

3

77

David Manheim

@davidmanheim

6 months

@DavidLammy @b_judah We absolutely should be addressing the crisis in Gaza, and ensure they have humanitarian aid. (The ceasefire being negotiated seems like a key step.) But UNRWA is part of the problem - it's structurally incompatible with both Palestinian and UN goals.

David Manheim

@davidmanheim

6 months

UNRWA is a failed model for providing assistance. If you want Palestinians to have a state, to have peace, to have any real government, you need to stop funding an organization that displaces government services and which gets funded only as long as there's no peace.

4

5

22

30

5

79

David Manheim

@davidmanheim

3 years

@kateferguson4 @robertwiblin In a battle between the fallacy of generalizing from fictional evidence vs. the idiocy of not extrapolating from obvious trends, it turns out the former helped us out, at least this one time.

Kate Ferguson

@kateferguson4

3 years

Matt Hancock says he overruled advice and ordered 100m doses of the Oxford vaccine rather than 30m because the film Contagion taught him there would be a global scramble for vaccine. He says every Brit needed to be jabbed “and I wasn’t going to settle for less” #LBC

533

596

4K

0

8

80