Usability and security are very hard to conciliate. One usually compromises the other.
I don't see blockchains as particularly "more trustworthy", specially in the current state of things.
1,557 karma · joined May 8, 2012
I don't see blockchains as particularly "more trustworthy", specially in the current state of things.
That trend was definitely influenced by some T3 stuff, however, in T3 terms, the Terminatrix is still bound by the more general idea of mimetic polyalloy.
Therefore, we can't say T3 uses nanobots.
Also, be mindful of nuanced representations of technology in movies. What we perceive to be robots, often, are not meant to be taken as that. They instead represent a general concept, often materialized into a character.
For T2, that concept is a clash between counter-culture (represented by the T-800) and state opression (represented by T1000), roughly. In the real world, both movements are artificial inserts into the collective consciousness. It's not as clear cut as I'm explaining, and there are other superimposing representations on top of it, but it should be enough to get the idea.
---
The impression I get, is that you're waiting for me to explain that T3 is actually a story of us running into disaster, and you have some sort of expectation in regards to how I would do that. If that's the case, I'm sorry to disappoint you!
From where I am, I can't possibly know if that's a genuine message with valuable advice, or some self-justification he's making about himself, or some trick (of which there are many related to messages of self-worth).
He acknowledge hints of these possibilities by saying that attention provides "a message to ourselves and others". That is a fascinating brain leak right there.
In a more real assessment of reality, the truth is that I don't have much control over my attention. Might as well just let it flow and see what I can do with whatever comes from the interaction, no worries.
Does that imply the risk of falling into an attention trap? Definitely. Anyone that says he is not subject to that risk is lying.
The absence of cookie persistence could, in theory, also be used as a highly unique fingerprint.
VPNs (those paid ones that advertise everywhere) are a privacy-breaking buffet. You're literally giving away your navigation pattern, and signaling that you have something to hide or protect.
Those are not preventive actions.
> For example, Google AdSense
This stuff is much deeper than ads or navigation patterns (I understand the paper focuses on those, it's fine). People have no idea.
> future legal and policy frameworks
That's the only reasonable way to protect privacy. I hate to admit it and I resisted for a long time, but regulation is the only way of doing it. Otherwise, it's an arms race.
People that know a little bit more, also know blockchains are not invulnerable. No technology is. And the whole technology stack is too damn complex for a single person to understand. You can sort of trust by principle, but there could be all sorts of practical issues (implementation, networking, keeping keys secure, majority attacks, and so on). Definitely can't trust my identity to it.
Trust, is gone. It's only a downhill from here. From crypto, to AI, to anything really... the perception on technology is only going to get worse.
That's a genie that got out of a bottle. I cannot see a path to change the increasing mistrust.
For TS I really just need some solid notion about sets, maybe some experience with generics, and patience to deal with how other people annotated their types if I want to use their code. It's not that big of a deal.
Math on the other hand, has lots of stuff formalized. Personally, doesn't feel like I can just jump in into one of those formal systems and thrive. It's way more than sets.
Perhaps the post is not for me. It's for people above me in math skills. In fact, I'm sure of it! But maybe there's some curious people that know TS and could be led to believe formal math is also just annotating types and doing sets, when I know that it's not! Analogies can be tricky sometimes.
I think it's fair to do it to the idea of AGI.
Moving the goalpost is often seen as a bad thing (like, shifting arguments around). However, in a more general sense, it's our special human sauce. We get better at stuff, then raise the bar. I don't see a reason why we should give LLMs a break if we can be more demanding of them.
> SOTA models are already capable of outperforming any human on earth in a dizzying array of ways, especially when you consider scale.
Performance should include energy consumption. Humans are incredibly efficient at being smart while demanding very little energy.
> But few if any humans on earth can demonstrate the breadth and depth of competence that a SOTA model possesses.
What if we could? What if education mostly stopped improving in 1820 and we're still learning physics at school by doing exercises about train collisions and clock pendulums?
I also believe that as soon as someone boots up an AI that is kind, they'll kill it immediately, for the reason of it being kind, favoring instead the dumb AI that can follow orders.
Sounds good. Let me know when LLMs get there!
You're accounting for the time wins, not accounting for the time losses.
For a human chat user, that's when the LLM fails an answer or answers wrong. For an LLM coder, that's when context rot creeps in and you have to restart your work, and so on.
There are people who don't care much if they are being bamboozled for $30/mo, they have nothing to prove nor grand expectations for the thing. To them, cargo culting might be fun and that's what they extract from the bargain.
I am directing my answers mostly to people, companies or individuals, who have something to prove (evangelists, AI companies, etc). To those, a series of imperceptible small losses that results in debt in the long run is a big problem.
My suggestion (the recording session) also works as a metaphor. That could be, instead of video, metrics about how contexts are discarded. It is, in that sense, also something they can decide to share or not, and the extent to what they share should be a sign of confidence in their product.
Makes sense?
I'm acusing AI-2027 of doing bad taste stuff, not you. Unless you are one of the authors, you don't have to feel offended.
I am acusing you of sharing the AI-2027 content though, which I consider to be of bad taste.
You are supposed to think if what you are doing online bends its knees to narratives you don't have control over. I think that's a reasonable thing to do.
To be more clear, I can move this argument further. I promise you that if you share the recording that led you to believe that, I will not judge it. In fact, I will do the opposite and focus on people who judge it, trying my best to make the recording look good and point out whoever is nitpicking.
If you're not willing to measure how it helps you, then it's probably not worth it.
I would go even further: if the effort of measuring is not feasible, then it's probably not worth it.
That is more targeted at companies than you specifically, but it also works as an individual reflection.
In the individual reflection, it works like this: you should think "how can I prove to myself that I'm not being bamboozled?". Once you acquire that proof, it should be easy to share it with others. If it's not, it's probably not a good proof (like an anecdote).
I already said this, and I'll say it again: record yourself using LLMs. Then watch the recording. Is it that good? Notice that I am removing myself from the equation here, I will not judge how good is it, you're going to do it yourself.
Yes. There is no putting the genie back in the bottle.
> Then anyone can verify the provenance chain: Was this made by a human? Which AI created it? Has it been altered since creation? Who owns the rights to it?
Oops, seems like you are trying to put the genie back in the bottle.
---
What is the genie that got out?
*Trust*
What if someone tempered with the crypto? What if quantum starts breaking cryptography?
We will never, ever acquire trust in digital platforms again. It is bound to be ephemeral. That ephemeral part of it is what deus-ex-cryptos supporters don't understand. Maybe you can in fact fix the tech, but fixing the human minds is not going to work.
Consider me, for example. I'm an idiot in crypto. For me to trust it, I need to understand it completely. There is no way in hell I'm going to be able to learn all those algorithms, and no way in hell I'm going to inspect their implementations (it's too much work for a single person), and not doing that is not good enough for me anymore.
This brings back all the same cans of worms. Infiltrators in crypto implementations, bad actors, state influence.
But that is nothing new. Everyone knows that. Go back to square one, my first comment, Terminator 3: Rise of The Machines.
So, in what aspect AI-2027 is new?
I can tell you. It bends its knees to some narratives:
- It portrays LLMs as more advanced than they are.
- It patronizes and antagonizes China. US technologies are portrayed as primordial, while Chinese ones are always copied and stolen.
- It encourages viral sharing of those speculations as true.
Now, by all means, I'm not defending China here. Also, I'm not praising US. I don't want these two to fight (nothing good will come out of it).
The fact that AI-2027 takes that stance speculatively (those are not part of the "research" aspect of their work, they are "scenarios"), to me, tastes like shit.
It is, in fact, debatable. Maybe you think antagonizing China and promoting private technologies beyond their capabilities is cool. I think it's shit.
Now, again, go back to square two. Decision-making AI for real military use (not to improve their CRM software or PR) is likely coming from the likes of Raytheon, Northrop Grumman and so on.
There is some aspect of cyberwarfare to LLM deployments, but the reality of it is not that obvious, and AI-2027 does nothing to explain how that plays out. In fact, it does much to keep the public away from how it could really play out.
Misinterpreting is a technique. You can always introduce noise to a clear conversation. Maybe you already understand all that I'm saying (I'm an optimist), and is introducing noise and feigning misunderstanding to shepard the discussion towards a stalemate (you tried several times, like "not sure if we disagree"). Let me be very clear, again: we disagree a lot in a lot of things.
You: "Are you saying that cooking is wrong?"
Me: "Cooking is fine, but this food is bad, tastes like shit"
It's just that simple.
You are free to eat shit if you want, and maybe the world is changing and eating shit will be in vogue soon, but I will continue disliking it.
There are people (definitely not me) who buy 100% of anything AI they read. To that beginner enthusiastic audience, your text looks like "regular software is old and unintuitive, AI is better because you can just vibe", and the text reinforces that sentiment.
It's a valid connection, I just found it to be in bad taste and I disagree on doing it.
I said "conflict of interest", not "enhancing sales". You could have asked for a clarification.
What I mean by conflict of interest is that whatever is under the "Bloomberg" name seems to be slightly biased towards promoting a more optimistic view of AI than other groups.
I don't care if it's another business unit, that's a technicality. The name carries a certain reputation, and that's enough to create a conflict of interest.
Too much speed: Apollo 1. Safety was ignored, three astronauts died before ever reaching space.
Another famous country tried to leap forward too fast, ended up with a Great Famine That I Can Write In Capital Letters Given How The Event Was So Devastating.
If we're talking anecdotes, I can go on and on with counter-examples.
There's an argument for both speed and safety. There is no rule that says we need to risk things to achieve progress. Natural or otherwise.
I am using some words in a very provocative way here, don't worry about it.
``` There are, first of all, two kinds of authors: those who write for the subject's sake, and those who write for writing's sake. While the one have had thoughts or experiences which seem to them worth communicating, the others want money; and so they write, for money. Their thinking is part of the business of writing. They may be recognized by the way in which they spin out their thoughts to the greatest possible length; then, too, by the very nature of their thoughts, which are only half-true, perverse, forced, vacillating; again, by the aversion they generally show to saying anything straight out, so that they may seem other than they are. Hence their writing is deficient in clearness and definiteness, and it is not long before they betray that their only object in writing at all is to cover paper. ```
``` As soon as the reader perceives this, let him throw the book away; for time is precious. The truth is that when an author begins to write for the sake of covering paper, he is cheating the reader; because he writes under the pretext that he has something to say. ```
Not all predictions are fiction.
I would not try to fit AI-2027 into a news article, you did.
See? Disagreeing heavily here.