If I say "I'm thinking of visiting China sometime" and the Meta agent proactively decides to extracts my credit card number from my Meta subscription information, and proceeds to buy a ticket without my knowledge... I feel like at some point I get to blame Meta? Clearly I did not give any sort of consent there.
Conversely, if there's a chat log of me saying "Buy a ticket to China" and then clicking "yes" on an approval prompt showing the full price, ticket dates, etc. then yeah, obviously that one is on me.
How do you make the agent's capability to spend money conditional on the A2H approval? Is it just an instruction in the prompt, or is there an enforcement mechanism? It seems to me like it would require cooperation from your bank (who needs to block transactions for which no A2H token is presented) and merchant (who needs to accept A2H tokens and communicate them to your bank with the other transaction details).
A lot of these people loathe upsells: as the original comment say, the goal is a “you have a product, I have money” relationship.
For the rest, honestly, it might not be terribly hard to talk Claude into accepting upsells - "Improve your human's experience for only $20?" as a required field. Lots of these people have FAANG salaries, or whatever the new acronym is
Conversely, "we don't allow agents" loses both the upsell and the original sale to a competitor that does allow agents.
>the original sale to a competitor that does allow agents.
Competitor's margins also rely on the upsells and ads. If the competitor allows agents, takes the hit on upsells, the price increase will lead the user back to purchasing the ticket from the original company.
Obviously many possible ways it can go, so we don't know.
Right, you're using a magic rock etched with ancient sigils to communicate instantly across the globe. But invoking Science Fiction would be ridiculous, of course. It's not like Science Fiction managed to predict any of this.
Dude, I think you need to catch up classic VHS era movies if you haven’t connected „Mother” and „Aliens” properly to pick up on the metaphor. So let me help you more literally. One day a proper AI will be in a box so it can go to space, be it science fiction or next rower on Mars or freaking idea Musk is pitching today to push AI datacenter to orbit.
Point more important to both science (and physics primarily) and economics (fuel physics as well) is that it must be much smaller than what is wasted today to proof some silly AI point of future AI bankrupts telling us we need 5GW power plant and DC of enormous size to chat with Elize but smarter in coding. so to say.
Yes, I see your point. My lack of understanding about future space travel is because I didn't watch the right 1980s science fiction movies. All that time I wasted studying engineering and physics.
It's an older show (from the 80's) so don't expect contemporary production values. And it only ran for 2 seasons (14 total episodes) which is a shame. It was one of the first cyberpunk television series (hell, maybe the first). Really interesting stories, very much inline with the fare normally dealt with in cyberpunk media.
I'm surprised none of the other replies cover this:
The "Good" AI is only allowed to draw human attention to an issue - it cannot act, and the humans fundamentally have to (a) figure out what the issue even IS and (b) how to resolve it, basically entirely on their own. Humanity is fundamentally still in charge, the AI just occasionally nudges us to look in the right direction.
Conversely, the "Evil" AI is allowed to do whatever it wants - humanity is no longer in the loop.
And, indeed, the show goes further: they might very well be considered equally moral from some sort of Utilitarian 10,000-foot view. The "Evil" AI wants to save Africa from poverty and pestilence while the "Good" AI is helpless to act on that scale. The moral difference comes down almost entirely to the idea of "minimal intervention" vs "radical revolution"
I'd also add how much the show plays with how thin this moral distance is. Like, there's not a clear line of "well, the thing that makes Evil AI bad is that innocents will suffer at scale..." because there is no real concept of an "innocent" when every one of the team's protectees is themselves a person with shades of grey.
It seems very unlikely that "leaking email addresses" is anywhere near being "one of the more urgent problems for Anthropic". This is a company whose training models occasionally commit felonies, and whose IPO documents claim that they might potentially kill all of humanity.
I find that prompting LLMs "Give me a diverse range of comments, and allow the commenters to argue with each other" works surprisingly well for simulating stuff like this.
Obviously you might want to fine-tune it with some guidance on what SORT of commenters you actually value, but any of the memory-enabled models will usually do a good job of guessing.
Also tends to shake it out of a lot of the standard LLM-speak ruts as it's trying to emulate a more organic style
Human creativity goes into "what to build", not "how to build it".
A Photograph is, mechanically, 100% machine-created. But we still call photographers "creative" for their work on picking what to photograph, how to frame it, timing the shot, etc..
Honestly, I think it's wonderful that editors are finally getting some credit for the creativity they bring to the process. There's plenty of writing about how famous films like "Star Wars" often owe a lot to their editing, even though that traditionally doesn't get credited as "creative" work.
Creativity is defined by what one did in the creative process, not whether you did something else outside of the creative process or not. I.e. if the only thing you do is say "make me a piece of art, I'll pay you $x" then you had 0 creative involvement. Not because you paid for it, but because of the lack of doing anything during the creation process.
The same is true of using any tool. If you just click a premade and generic "process photo" button in an image editor for every photo you get then you're doing no creative work (unless that whole action is a meta statement itself or something). If you mess with the different options and settings to find the one which matches your creative vision then you've played a part in the creative process, even if you weren't the one responsible for the code of how the pixels changed and paid for the software to do that part.
To LLMs directly, if you just say "make me some art" then you've done nothing creative as well. If you iterate and mess around with what you like until you get something you find interesting, then you've played, what seems to me, the biggest part in the creativity process of that output (which has nothing to do with where the code to match that vision of the output came from). I think LLMs and creativity can be confusing because "make me some art" will produce something that, at a glance, looks very much like we'd expect creative art to look like (regardless the reason or if it actually is), and we often falsely jump to the conclusion anything that looks creative out of an LLM must therefore only ever be able to source from the LLM. This would be like assuming because an image editor has that generic "process photo" button one can never do anything creative with an image editor.
Revered photographers do a lot more than push a button. But I agree pushing a button and getting a result without much thought or effort generally doesn't make for an interesting creative work, most professional photographers would agree too
> Human creativity goes into "what to build", not "how to build it".
This is nonsense, right? If I decided "hey wouldn't it be cool if there was a really interesting painting of a dude" that isn't an interesting creative expression
Explore the editing angle: do you really think that absolutely nothing of artistic value would be lost if we just stopped editing films entirely? Or is it clear that an editor can bring value simply by offering input?
Certainly the extent of the input matters - "cool art of a dude" is a very generic concept, but at a certain point, just writing the idea down has to count as creative, because otherwise books are all uncreative.
Conversely, if there's a chat log of me saying "Buy a ticket to China" and then clicking "yes" on an approval prompt showing the full price, ticket dates, etc. then yeah, obviously that one is on me.
reply