theptip

50 ms·

871.
by theptip 3y ago
Tail risk / volatility.
872.
by theptip 3y ago
Author rediscovers Kanban and Lean! “Limit WIP” is a well-known recommendation in agile circles. It’s great except for the situations where it doesn’t work, eg. you have multiple customers with must-do features, you have support commitments
873.
by theptip 3y ago
Every modern backend web programming framework? Frontend requires a lot of flexibility and compromise because the runtime environment is so disparate. In the backend you choose and control one runtime so the framework can integrate tightly
874.
by theptip 3y ago
Exactly. You don’t want to be bleeding edge (churn, bugs) but in general you usually don’t want to be on the oldest supported version either (let alone unsupported). Risk/reward depends on the usecase of course. For a startup I’d be on
875.
by theptip 3y ago
I think you can envision “non-human-like”. Eg we already imagine animals can be conscious. Conscious just means an intelligence that is self-aware and has subjective experience. If you define that as human-like then your point stands tautol
876.
by theptip 3y ago
People tend to have strong opinions about this sort of thing, but the fact of the matter is there’s no way (currently) to know. (But also, it’s effectively metaphysics as it makes no testable/falsifiable predictions.) We could in princ
877.
by theptip 3y ago
I wonder if there are tools that library maintainers can use to calculate transitive code coverage of their libraries, eg which other lib test suites use them? Could be an interesting test strategy, in some sense it’s directed fuzz testing.
878.
by theptip 3y ago
Git is a lot less scary when you understand the reflog; commit or stash your local changes and then you can rebase without fear of losing anything. (As a bonus tip, place “mybranch.bak” branches as pointers to your pre-rebase commit sha to
879.
A Big Week in Medical A.I(erictopol.substack.com)
2 points by theptip 3y ago|0 comments
880.
by theptip 3y ago
> Except, for Google, I did not use the "how do I" etc. and just threw some keywords at it: "python f-string wrapping", "python f-string folding", etc.
881.
by theptip 3y ago
Presumably you don’t publish your strats there.
882.
by theptip 3y ago
Semi analysis posted recently noting that Meta locked in these purchases a while ago; something like a year or more. So they probably didn’t pay today’s spot rate.
883.
by theptip 3y ago
The kind AIs will just sit and meditate, or organize your calendar. The unkind AIs will seek power, and in the limit will tend to dominate. There is no stable attractor around the “be kind” strategy. (Also remember that a smarter-than-you A
884.
by theptip 3y ago
No, they are not. Pretty much everyone in the x-risk community also recognizes the existence of short-term mundane harms as well. The community has been making these predictions for over a decade, long before it was anything other than craz
885.
by theptip 3y ago
> No one's trying to eliminate anyone Probably not any consolation to the Neanderthal or any other animal species that is out-competed in nature. Doesn’t have to be malevolent or intentional for a more capable species to render oth
886.
by theptip 3y ago
Personally I think the “foom” scenarios with very short timelines are unrealistic. However I think the line of argument in the parent comment is extremely weak. At some point the super-intelligence can build its own actuators; for example h
887.
by theptip 3y ago
“01.ai” is not a very auspicious name; 01 was the first AI nation, that eventually waged war with humanity and then enslaved them in The Matrix.
888.
by theptip 3y ago
Spending time and energy to engage in reading a book can change the way you think, and subsequently change your interpretation of the text. A good book is not just a bag of facts, but a tool to change your cognitive processes.
889.
by theptip 3y ago
I stand corrected, as stated my claim about charity OOM was wrong. Still, I don’t think I need to update much. Because: > So, offer a return on investment This is precisely what they did; Microsoft’s investment is an extremely funky capp
890.
by theptip 3y ago
This would be orders of magnitude more than any charity has ever raised, and is also an uncommonly huge raise even among _for-profit_ companies where investors expect returns. Even when the EA community was flush with crypto billionaires th
891.
by theptip 3y ago
There is no way OpenAI could have raised $10B as a non-profit.
892.
by theptip 3y ago
A bit of a cherry-pick, no? You can’t just look at one bad thing that happened, you need to weigh the good and bad that comes with a particular policy stance. Concretely in this case it seems quite relevant to include GDPR, which consumers
893.
by theptip 3y ago
PID controllers are for linear or mildly-nonlinear systems. Plasma fluid dynamics are about as nonlinear as it gets.
894.
by theptip 3y ago
This seems like spurious reasoning. It’s like saying “rockets are 1950s technology therefore SpaceX is going to fail”. Commonwealth are using cutting-edge high-temp superconductors which can generate much stronger fields: https:/&#x
895.
by theptip 3y ago
I think the DeepMind taxonomy is good; recognize that AGI is too broad a term, so add levels to make it specific. https://arxiv.org/abs/2311.02462 “Competent AGI” (or a little earlier) would be my guess for where OpenA
896.
by theptip 3y ago
I don’t think this belief was widespread at all at that time. Indeed, it’s not widespread even now, lots of folks round here are still confused by “open weight sounds like open source and we like open source”, and Elon is still charging tow
897.
by theptip 3y ago
An interesting nugget, Musk is here claiming that GPT-4 is already AGI, and therefore not covered by the MS licensing deal.
898.
by theptip 3y ago
By the same token, you are asking people to introspect and unilaterally sacrifice in a way that, for the same reason, is not going to happen for pretty obvious reasons. I’m on board with solving this at the governmental level FWIW. I suppor
899.
by theptip 3y ago
It’s easy to defend this position. It is safer to operate an AI in a centralized service, because if you discover dangerous capabilities you can turn it off or mitigate them. If you open-weight the model, if dangerous capabilities are later
900.
by theptip 3y ago
I don’t think that’s true. Google took their ad money and placed a famous number of bets on moonshots, more than any company in a long time (back as far as Bell Labs? Did even they fund their research to the extent Google has?) The whole Me

More ›