But you need to verify everything unless it’s self evident. The number of times CoPilot (Sonnett 4) still hallucinates Browser APIs is astonishing. Imaging trying to learn something that can’t be checked easily, like Egyptian archeology or something.
You have to verify everything from human developers too. They hallucinate APIs when they try to write code from memory. So we have:
- documentation
- design reviews
- type systems
- code review
- unit tests
- continuous integration
- integration testing
- Q&A process
- etc.
It turns out when include all these processes, teams of error-prone human developers can produce complex working software. Mostly -- sometimes there are bugs. Kind of a lot actually. But we get things done.
Is it not the same with AI? With the right processes you can get consistent results from inconsistent tools.
Taking the example of egyptian archeology, if you're reading the work of someone who is well regarded as an expert in the field, you can trust their word a lot more than you can trust the word of an AI, even if the AI is provided the text you're reading.
This is a pretty massive difference between the two, and your narrative is part of why AI is proving to be so harmful for education in general. Delusional dreamers and greedy CEOs talking about AI being able to do "PhD level work" have potentially ruined a significant chunk of the next generation into thinking they are genuinely learning from asking AI "a few questions" and taking the answers at face value instead of struggling through the material to build true understanding.
There needs to be a reasonable chance of correctness. At least the local toddlers around here don’t randomly provide a solution to a problem that would take me hours to find but only minutes to validate.
>I'll take a potential solution I can validate over no idea whatsoever of my own any day.
If you have to validate what the LLM says, I assume you'd do that by researching primary sources and works by other experts. At that point, the LLM did nothing except charge you for a few tokens before you went down the usual research path. I could see LLMs being good for providing an outline of what you'd need to research, which is definitely helpful but not in a singularity way.
> If you have to validate what the LLM says, I assume you'd do that by researching primary sources and works by other experts.
For research, yes, and the utility there is a bit more limited. They’re still great at digesting and contextualizing dozens or hundreds of sources in a few minutes which would take me hours.
But what I mean by “easily testable” is usually writing code. If I already have good failing tests, verification is indeed very very cheap. (Essentially boils down to checking if the LLM hacked around the test cases or even deleted some.)
> At that point, the LLM did nothing […]
I’d pay actual money for a junior dev or research assistant capable of reading, summarizing, and coming up with proofs of concept at any hour of the day without getting bored at the level of current LLMs, but I’ve got the feeling $20/month wouldn’t be appealing to most candidates.
All of the information available from an LLM (and probably more) is available in books or published on the internet. They can go to a library and a read a book. They can be fairly certain books written by subject matter experts aren’t just made up.
Sure, I just gave the Browser API example as evidence that the 'hallucination' problem is not gone.
OP said it's like "talking to a professor" and you can use it to learn college level stuff. This is where I disagree. I did not double check my professors or text books usually.
I can't reproduce the "worst case" i.e. that if I ping "google.com" it gets first sent to "google.com.fritz.box". I'm on windows 10 and have a FRITZ!Box 7590.
If I ping "google.com" It just queries google.com
If I ping a domain I have never visited it just does a query for that domain.
It only appends .fritz.box if I, e.g. only ping "google".
So maybe they fixed it? I also changed quite a lot of settings throughout the years.
HKEY_LOCAL_MACHINE\Software\Policies\Microsoft\Windows NT\DNSClient\AppendToMultiLabelName
[default value - 0 (Do not Append Suffix)]
which resolver built into Windows (DNS Client) respects.
nslookup contains its own DNS resolver and does not rely on the resolver built into the operating system. The DNS (multi-label) query packets sent by the nslookup tool will append the domains listed in the suffix search order (or primary DNS suffix if the list is empty) irrespective of that registry key.
In summary, don't use nslookup to try to get insight into what actually happens when apps/services try to resolve names. ping is probably a better bet, at least it uses Windows resolver which honours the above registry key.
I think I agree, if that is legally possible.
I don't think apple provides any important infrastructure like Amazon, Microsoft and Google do.
It's basically a luxury company.
Might be bad for the people who already own iOS devices and need it to survive in one way or another (e.g. as a wallet).
Well apple could still operate the devices that they've sold, they just wouldn't be able to sell for a specified period of time, doubtful apple removes support anyways because then they customer trust.
As plenty of others have said - nope, dissaving is very uncommon word. As a native American English speaker, I don't recall ever seeing it either.
One clue for future reference - the linked article had to clarify the meaning of the word _in the title_. That strongly indicates to me that, even if in certain areas the word is well understood, the author expected there to be confusion.
Battery life was about a week and is now about 3 days but it really depends on which apps you want to "freeze" and which apps you never want to "freeze" (to receive notifications).
There’s also a "super saving mode" which allows only 6 apps, no notifications. I use it during long bikepacking trips, recharging every other day when stoping for lunch in a restaurant. It is great. Also, I leave the "super saving mode" if I need to use another app.
So it is perfectly possible to get very long battery time but you need to really configure it and figure out how to use it.
I’ve lost a lot a battery time by disabling freeze for Adguard VPN. In day to day use, I prefer to use more battery and send less data to China (the stuff is full of spyware). Even with that, I charge it every three days but I don’t use it much (which is the goal)
Everytime I read about longevity research and how many people are in favor of it I can't stop thinking about this speech.
And one of the endings of Cyberpunk 2077.
I think people generally, and the Silicon Valley set in particular, have a hard time abstracting from “would I like” to “would the world be a better place if”.
Would I like to live a thousand years? Yes, with the obvious caveats.
Would the world be a better place if the technology for living a thousand years existed? Absolutely not, at least not at first, and certainly not today. There’s a great many people around right now who’s primary redeeming quality is their impending mortality - it’s not just science that advances one funeral at a time.
How many times has your life or someone close to you in your life not died from something they would have died of 100yrs ago? If you're happy medical tech saved their lives then you're arguably for life extension because all it really means is saving more lives from more things that kill them.
I don’t know if this is supposed to be a dunk or something, but - yes, my grandma lived to 93 because of modern medicine. I was happy she did. That’s the tension: things that are good for me personally can be bad for the world at large (I mean, not my grandma’s longevity specifically - she was a lovely woman), and a big part of emotional and intellectual maturity is recognizing that indeed the world is full of tradeoffs and I can’t have everything I want.
Specific to:
> all it really means is saving more lives from more things that kill them.
No, that’s not all it really means, not in our society, not in our time. As Ted Chiang put it, “Most of our fears or anxieties about technology are best understood as fears or anxiety about how capitalism will use technology against us,” and that’s also the case here: the outcome of this technology isn’t that my grandma lives to 150, it’s that Vladimir Putin lives to 150. If my grandma needs to die at 90 so we don’t have immortal god-emperors - if I have to die at 90 - then so be it. Some day we may live in a world where longevity technology is an unalloyed good, but until that day, we don’t get to just put the good stuff on the ledger and ignore the bad stuff.
Most notably Rupert Murdoch- while I do not wish death on the man, it's certainly true that he has a grip on the hearts and minds of people and often uses his media empire to convince people to go against their own interests.
He will be replaced by someone similar, but seldom are people as effective as their predecessors.
The fact that you would even make this joke shows how absurdly far we have fallen.
Obviously they exist for that purpose, studying the foundations of news media and journalism... for even a day... shows concisely that it was painfully created for this reason.
> He will be replaced by someone similar, but seldom are people as effective as their predecessors.
I'm not a real believer in the "Great Man" theory of history - I think the ground needs to be set for an event for it to happen, I don't think the will of one person is truly sufficient to bend history - but there are certain people who you would have a very, very hard time replacing in a given scenario.
Rupert Murdoch is definitely one, and Donald Trump is another - without getting into specific judgements of the man, there's nobody else within easy reach who could do what he's done, and I don't really see his movement surviving him. He's a particular person for a particular moment, and it's hard to see anyone else doing what he has.
“All men have an emotion to kill; when they strongly dislike some one they involuntarily wish he was dead. I have never killed any one, but I have read some obituary notices with great satisfaction.” – Clarence Darrow
“a new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents die, and a new generation grows up that is familiar with it.” – Max Planck
1) Possible solutions aren't binary (true vs false) but trinary (true vs false vs indeterminate)
2) The devil is always in the details. The world is fucking complex and and a first order approximation isn't going to get you there anymore. We've had 100kyrs to solve problems, we got most of the simple ones down (appearing simple does not mean simple)
2.5) A clique wouldn't be a clique if it wasn't something practically everyone knows and can recite but is not something people demonstrate an actual understanding of by observing their actions. (Just like LLMs: just because you can repeat some knowledge does not mean you're able to (ineptitude), or have the will to (malice), use the knowledge in any meaningful way)
> Would the world be a better place if the technology for living a thousand years existed? Absolutely not, at least not at first, and certainly not today
If you want to sacrifice your life for a better world, that is your decision. But do not force that decision to other people.
We could live for 20 years or 200 and it wouldn't matter - entities will emerge that will attempt to consolidate and abuse power. Those may be individual dictators, tyrannical governments, or global conglomerates. The answer is the same, and it doesn't involve hampering scientific progress.
