Richard121

Karma: 46

Richard121 Jan 26, 2025, 8:40 AM
−5 points
0
in reply to: Matt Levinson’s comment on: The Case Against AI Control Research
“Corporate leaders might be greedy, hubristic, and/or reckless”
They are or will be, with a probability greater than 99%. These characteristics are generally rewarded in the upper levels of the corporate world—even a reckless bad bet is rarely career-ending above a certain level of personal wealth.
The minimum buy-in cost to create have a reasonable chance of creating some form of AGI is at least $10-100 billion dollars in today’s money. This is a pretty stupendous amount of money, only someone who is exceedingly greedy, hubristic and/or reckless is likely to amass control of such a large amount of resources and be able to direct it towards any particular goal.
If you look at the world today, there are perhaps 200 people with control over that much resource, and only perhaps only ten for whom doing so would not be somewhat reckless. You can no doubt name them, and their greed and/or hubristic nature is well publicised.
This is not the actual research team of course, but it is the environment in which they are working.

Richard121 Dec 28, 2024, 8:43 AM
−2 points
−7
in reply to: Milli | Martin’s comment on: When Is Insurance Worth It?
If you already know that an adverse event is highly likely for your specific circumstances, then it is likely that the insurer will refuse to pay out for not disclosing “material information”—a breach of contract.

Loss leaders are common, but rely on a reasonably accurate prediction of what that loss will be. and simple errors are of course common.

I think “nobody knows what they’re doing” is an overstatement. Insurance companies make significant profit, and very rarely fail. The usual complaint I see from investors is that some years an insurer has made less profit than others—which is exactly what one should expect, so clearly these investors don’t understand what they’ve invested in.

Your bike example sounds like bad luck. An event with a probability of 1% per year can still happen three years running.

It would be interesting to consider the inverse Kelly—given an insurance premium X, what are the reasonable bounds of assumed probability of an adverse event costing Y?

I’m not sure how the wealth of the underwriter could be estimated though.

Richard121 Dec 24, 2024, 11:23 AM
1 point
0
in reply to: Milli | Martin’s comment on: When Is Insurance Worth It?
In the majority of cases the insurer has more knowledge, as they can afford the time to do the research.

This is of course why any insurance market must have a great many suppliers. If there are only one or two insurers, they can easily set the premiums to make supernormal profits because they are ‘competing’ against consumers with very little knowledge, instead of other insurers who have similar knowledge levels.

Richard121 Dec 24, 2024, 11:14 AM
2 points
1
in reply to: jmh’s comment on: When Is Insurance Worth It?
Indeed. In reality, the vast majority of people do not have sufficient information to make reasonable estimates of the probability of loss—and in many cases, even the size of the loss.

Eg a landlord is required to rebuild the place and temporarily rehome the tenants while that’s being done in the event of the house being destroyed by fire or flood. They’re also liable for compensation and healthcare of affected third parties—and legal cost of determining those figures.

They can probably calculate the rebuilding cost and a reasonable upper bound on temporary housing. But third party liability?

So, it still comes back to vibes.

Richard121 Dec 24, 2024, 10:57 AM
1 point
0
in reply to: kqr’s comment on: When Is Insurance Worth It?
A refused claim is (legally) an event that was never covered by the insurance, and is therefore irrelevant if the question is “take policy A or not at all”.

After all, if that event occurred without insurance, it is still not covered.

However, this is important to consider when comparing different policies with different amounts of coverage. Eg “comprehensive” car insurance compared with “third party, fire, and theft”.

“Rates” of unpaid claims only make sense in a situation where the law allows the insurer to breach their contracts. In that situation, the value of insurance plummets, and possibly reaches zero.

Richard121 Oct 23, 2024, 8:14 AM
6 points
0
in reply to: Stephen Fowler’s comment on: Arithmetic is an underrated world-modeling technology
For those who would like a hint.

In English, “And” generally indicates addition, “Per” division.

Now consider which of the following makes sense:

Ferrets and seconds

Ferrets per second

Richard121 Sep 14, 2024, 10:50 AM
1 point
1
in reply to: Morpheus’s comment on: My AI Model Delta Compared To Christiano
The issue there is that “best X” varies wildly depending on purpose, budget and usage.

Take a pen: For me, I mostly keep pens in my bag to make quick notes and lend out. The overriding concern is that the pens are very cheap, can be visually checked whether full or empty, and never leak, because they will spend a lot of time bouncing around in my bag, and I am unlikely to get them back when loaned.

A calligrapher has very different requirements.

Richard121 Sep 8, 2024, 9:48 PM
9 points
0
on: Would catching your AIs trying to escape convince AI developers to slow down or undeploy?
The really short answer:

No. The lab would not shut down. It would probably not even notify anyone outside the company of the alignment problem or the escape attempts.

The evidence would be buried, and is unlikely to come to the attention of anyone outside the lab until after a devastating incident.

For real-world examples, look to other industries.

Perhaps the two clearest examples are:
1. The UK Post Office “Horizon” computer system. This was known to be making egregious errors, yet postmasters were prosecuted and imprisoned on the known-faulty say-so of the Horizon system.
2. Grenfell Tower Fire. The public enquiry has found multiple cases where the evidence that the cladding was highly flammable and unsuited for use on tall buildings was deliberately hidden from the authorities, and even when the authorities knew (or should have known), the evidence was ignored.
Similar cases can be found throughout the EEA, US and Canada.

Richard121 Sep 8, 2024, 9:01 PM
7 points
1
on: Nursing doubts
The advice we were given was “Fed is best”.

However, we also very much wanted to breastfeed, primarily due to convenience and cost.

Getting started was very difficult. We used a small number of premixed formula bottles, two of which were “free samples”. We almost gave up.

A lactation consultant helped us find a comfortable position, and eventually ended up with the “rugby hold”, which makes no intuitive sense whatsoever—the infant is held under the arm, legs almost behind the mother.

Once started, breastfeeding turned out to have several clear advantages, some of which were very surprising to us:
1. Low poo volume. Breastfed babies produce very little poo early on, as almost everything is absorbed. Formula produces a lot more poo. (This is apparently well-known among midwifes & health visitors, but not mentioned until we actively asked)
2. Low to no poo stink compared to formula. (Anecdotal) In discussion with other parents we found that the formula fed infants poo smelled a lot worse, even accounting for volume and parental preference (your own child’s poo stinks less than that of all other children)
3. Reduced rate of constipation. Formula fed infants are at a much higher risk of constipation, as a side effect of the above.
4. Reduced posseting. Bottle fed infants swallow a lot more air due to the mechanics involved, which makes them need more burping and bring up more milk when they do. Obviously this is due to the bottle itself rather than what’s in it, but formula fed is 100% bottle while breastfed is some lower percentage.
5. Night feeds didn’t require full waking. The mother can do it half asleep and the infant is fed almost instantly, as mentioned elsewhere. However, this can only be done by the breastfeeding mother. The partner cannot assist.
6. (Anecdotal) Flying was easy. When disturbed, simply insert nipple and infant is immediately calmed. However it is possible a pacifier would work similarly well for a bottle fed infant.

Richard121 Jan 23, 2024, 11:00 PM
3 points
1
in reply to: Tao Lin’s comment on: There is way too much serendipity
Indeed. The incentives to put new ones on the market are very limited, due to legalities and economics.
A corporation has a limited window of patent monopoly, so a fairly short time period to recoup their investment to develop, licence and build out manufacturing capacity—thus need to sell it for a high price or sell high volumes.
It is a long and expensive process to get a new compound approved for use as a food additive, and it needs to be done separately in each major jurisdiction—at least China, USA and EU.
A new sweetener is directly competing against all the other ones that are already on the market—merely ‘being much sweeter’ isn’t enough.
It has to be significiantly better in some other way, if only because it will be considerably more expensive at first.

Richard121 Nov 11, 2023, 10:56 PM
3 points
0
in reply to: O O’s comment on: AI Timelines
I think there is a significant societal difference, because that last step is a lot bigger than the ones before.
In general, businesses tend to try to reduce headcount as people retire or leave, even if it means some workers have very little to do. Redundancies are expensive and take a long time—the larger they are, the longer it takes.
Businesses are also primarily staffed and run by humans who do not wish to lose their own jobs.
For a real-world example of a task that is already >99% automatable, consider real estate conveyancing.
The actual transaction is already entirely automated via simple algorithms—the database of land ownership is updated indicating the new owner, and the figures representing monetary wealth are updated in two or more bank accounts.
The work prior to that consists of identity confirmation, and document comprehension to find and raise possible issues that the buyer and seller need to be informed about.
All of this is already reasonably practicable with existing LLMs and image matching.
Have any conveyancing solicitors replaced all of their staff thusly?

Richard121 Nov 11, 2023, 10:19 PM
23 points
−8
on: AI Timelines
1000x energy consumption in 10-20 years is a really wild prediction, I would give it a <0.1% probability.
It’s several orders of magnitude faster than any previous multiple, and requires large amounts of physical infrastructure that takes a long time to construct.
1000x is a really, really big number.
Baseline
2022 figures, total worldwide consumption was 180 PWh/year^[1]
Of that:
- Oil: 53 PWh
- Coal: 45 PWh
- Gas: 40 PWh
- Hydro: 11 PWh
- Nuclear: 7 PWh
- Modern renewable: 13 PWh
- Traditional: 11 PWh
(2 sig fig because we’re talking about OOM here)
There has only been a x10 multiple in the last 100 years—humanity consumed approx. 18 PWh/year around 1920 or so (details are sketchy for obvious reasons).
Looking at doubling time, we have:
1800 (5653 TWh)
1890 (10684 TWh) − 90 years
1940 (22869 TWh) − 50
1960 (41814 TWh) − 20
1978 (85869 TWh) − 18
2018 (172514 TWh) − 40
So historically, the fastest rate of doubling has been 20 years.
Build it anyway
It takes 5-10 years for humans to build a medium to large size power plant, assuming no legal constraints.
AGI is very unlikely to be able to build an individual plant much faster, although it could build more at once.
Let’s ignore that and assume AGI can build instantly.
What’s in the power plant
At current consumption, known oil, gas and coal reserves are roughly 250 years in total.
Thus at 1000x consumption they are consumed in less than three months.
Nuclear fuel reserves are a similar size − 250 years of uranium, so assuming reprocessing etc, let’s say 1000-2000 years at 2022 consumption.
So the AGI has less than 3 years of known fuel reserves at 1000x current consumption.
However, “reserves” means we know where it is and how much could be economically extracted.
Exploration will find more, and of course there are many other, more esoteric methods of electricity generation known or believed to be possible but currently uneconomic or unknown how to build.
How about Space?
Solar irradiance is roughly 1380 W/m2 at Earth’s orbital distance. Call it 12 MWh/year/m2, or 12 TWh/year/km2
We’re looking for 180,000,000 TWh/year, so we need a solar panel area of around 20,000,000 km2 at >50% efficiency.
That’s a circle >2500km radius—much bigger than the Moon!
Fusion
The hidden assumption is that AGI not only figures out large-scale fusion in 4 years, but rolls it out immediately.
1. ^
  Hannah Ritchie, Pablo Rosado and Max Roser (2020) - “Energy Production and Consumption”
  https://ourworldindata.org/energy-production-consumption

Richard121 Nov 11, 2023, 9:06 PM
5 points
0
in reply to: Daniel Kokotajlo’s comment on: AI Timelines
I think the gap between AI R&D being 99% automatable and being actually automated will be approximately one day
That’s wildly optimistic. There aren’t any businesses that can change anywhere near that fast.
Even if they genuinely wanted to, the laws 99% of business are governed by mean that they genuinely can’t do that. The absolute minimum time for such radical change under most jurisdictions is roughly six months.
Looking at the history of step changes in industry/business such as the industrial and information revolutions, I think the minimum plausible time between “can be automated with reasonable accuracy” and “is actually automated” is roughly a decade (give or take five years), because the humans who would be ‘replaced’ will not go gently.
That is far faster than either of the previous revolutions though, and a lot faster than the vast majority of people are capable of adapting. Which would lead to Interesting Times...

Richard121 Oct 3, 2022, 11:59 PM
2 points
0
in reply to: artemium’s comment on: A Few Terrifying Facts About The Russo-Ukrainian War
The main reason for everything being in a crappy state is almost certainly (>90%) widespread corruption.
Everyone who can is creaming off a little bit, leaving very little for the actual materiél and training.
So shoddy materials, poor to no training, missing equipment, components and spares.
That said, while it is very likely that the Russian nuclear arsenal is in extremely poor state, and I’d possibly go as high as ⁵⁰⁄₅₀ that their ICBMs could launch but cannot be aimed (as that takes expensive components that are easy to steal/not deliver and hide that fact), missing the target by a hundred miles or more is basically irrelevant in the “ending the world” stakes.
A ‘tactical’ device doesn’t need much in the way of aiming, and on the assumption that it does in fact contain nuclear material there’s not a huge civilian difference between it exploding ‘as designed’ or “just” fizzling.
If only the initiator went off, the weapon disintegrated during launch/firing, or the weapon/aircraft was shot down, it would still spread radioactive material over a wide area.
While that wouldn’t be the “shock and awe” of a mushroom cloud, it’s still pretty devastating to normal life.

Richard121 Jan 1, 2022, 11:39 PM
2 points
in reply to: Zvi’s comment on: Omicron: My Current Model
It is absolutely certain that there will be more “variants of interest”.
This is basically the evolutionary modelling that pretty much all Governments have ignored, every time—Delta and Omicron were predicted by all eviolutionary biologists.
The open questions are:
- Whether there will be a new variant of interest that is notably more infectious, and thus becomes dominant after Omicron.
  In the UK, Delta completely outcompeted all other variants in around 3-4 months (>95% of all sequenced cases were Delta). Omicron is expected to do the same by Feb if not earlier. USA is likely similar, albeit delayed by a few weeks.
- Whether future variants cause more or less serious disease than Omicron.
- When this will occur.
  To me, it seems most likely this will be Feb/March 2022 or Fall 2022
If the answer to the first question is Yes, and the second question is “far less serious”, then the pandemic is over When it occurs—it has become another ‘common cold’ and is unlikely to mutate further to produce more serious disease (because it didn’t).
However, if it is Yes and The Same Or More Serious, then we will certainly need further booster jabs in Fall/Winter 2022, perhaps tailored more closely.

Richard121 Mar 4, 2020, 7:05 AM
5 points
in reply to: Davidmanheim’s comment on: Coronavirus: Justified Practical Advice Thread
In many ways COVID19 is irrelevant.

It’s already spreading in multiple countries, and is very likely to become endemic.

However: It is not alone, there will be future viruses.

We should be looking to create habits that will protect us from both COVID19 and all future local epidemics and pandemics.

That means habits that can be maintained indefinitely, not short-term changes that are not sustainable.

Things like washing hands and using hand sanitizer and creams are habits you can learn and maintain.

Checking everyone for fever, selling your stocks to buy X is not sustainable, and so these behaviours will be quickly forgotten.

Richard121 Mar 4, 2020, 7:00 AM
2 points
in reply to: FactorialCode’s comment on: Coronavirus: Justified Practical Advice Thread
Hand sanitizer is a poor substitute for actually washing your hands with soap and water.

Coronaviruses are “enveloped” viruses, which means they have a fat-based shell that protects the genetic material and (presumably) aids it in infecting a cell.

Destroying this shell “kills” the virus.

While an alcohol sanitizer can of course dissolve the fats in the shell, it is difficult to get enough alcohol all over the skin to do this.

Soap is more effective because it actively attacks fats, and of course washing your hands provides far more volume and time in which to destroy the virus shells.

Richard121 Mar 4, 2020, 1:07 AM
2 points
in reply to: zby’s comment on: Seeing the Smoke
Some data from the BBC comparing them: https://www.bbc.co.uk/news/world-51235105

The current overall death rate is estimated at approx. 1%, with a fairly large number of cases.

It has spread much further and faster than SARS or MERS, but is far less dangerous than either.

Which makes sense—a disease which incapacitates or kills in a high percentage of cases is unlikely to spread as fast as one which has mild symptoms in most infected people.

It now appears almost certain to become endemic. The real goal is to slow down the spread sufficiently to develop vaccines before this happens.

That worked for bird flu and swine flu. It remains to be seen for Covid19

Richard121 Mar 4, 2020, 12:57 AM
2 points
in reply to: Richard121’s comment on: Seeing the Smoke
Couple of numbers to think about:

A flu shot covering the four or five “most common” viruses (inc. bird flu) in 2019/2020 cost $10 privately in the UK, and $20-$75 in the USA.

In the UK, everyone registered with a GP who is at increased risk of pneumonia is offered the shot for free, regardless of status. https://www.nhs.uk/conditions/vaccinations/who-should-have-flu-vaccine/

In the USA, the Affordable Care Act made the flu shot free for most people who have health insurance, and for some groups without insurance. Approx. 10% of US citizens have no health insurance at all, these are of course those who either don’t work or have low-paid jobs.

Richard121 Mar 4, 2020, 12:43 AM
−3 points
on: Seeing the Smoke
If you live in a nation that has universal and free healthcare, then there is in fact very little reason to worry.

Wash your hands and practice good hygiene.
- As you should anyway, because that reduces the spread of other diseases inc. flu and colds, which are already endemic.
If you live in a nation like the USA, then you should worry, for several reasons:
1. Many workers cannot afford to take any time off, and cannot afford to go to their doctor for any treatment until it is already life threatening to them. By that time they are likely to have spread it to customers and colleagues.
2. Many people cannot afford immunizations. Their insurance doesn’t cover them, or their co-pay is high.
3. These workers also work physically closely with colleagues, and are more likely to travel on buses due to the cost.
Full-service restaurants are commonly extremely low pay (below minimum wage) with little to no health coverage. Yes, there are laws/regulations about food service, but they are routinely ignored—staff can’t afford time off and managers turn a blind eye.

There is currently no vaccine and no cure—only general support. So no amount of money will save you from death if you are one of the unfortunate few whose immune system cannot destroy the virus before multiple organ failure.

The group with the highest risk of death (~15%) from Covid19 are old men with heart problems, esp. if they have further co-morbidities.

We can thus predict that quite a few old, rich men are going to die because of the US healthcare system.

If Covid19 worries you, then campaign for free healthcare for all because it is the only way to protect yourself from it—and the future viruses that will inevitably follow.

Richard121

Baseline

Build it anyway

What’s in the power plant

How about Space?

Fusion