In my opinion the question isn’t so much “if” but rather “when”.
When will AI research and hardware capabilities reach a point that it’s practical to embed something like that into a regular document?
We’ve already seen proof of concept LLMs embedded into OpenType fonts.
I guess the other question is then “what capabilities would these AI agents have?” You’d hope just permission to present within that document. But that depends entirely on what unpatched vulnerabilities are lurking (such as the Microsoft ANSI RCE also featured on the HN front page)
> // Use interpreted JS only to avoid RWX pages in our address space. Also, --jitless implies --no-expose-wasm, which reduce exposure since no PDF should contain web assembly.
The first widespread AI Malware will be a historic moment in this century. It will adapt like a real biological virus to its host and we have no cure for this.
I once changed my schedule for a month to sleeping 4 hours, waking for 4, sleeping 4, and then waking for 12. I would work in the day, come home, sleep 4, wake 4, and then sleep 4 more, then off to work again. On weekends I switched to a normal 16/8 schedule.
I got plenty of rest and got all my domestic stuff done overnight - 24h grocery store, laundromat, etc. I was never unusually tired.
The down side is that it mostly killed my social life. So I quit; the advantages didn’t outweigh the disadvantage of being on a different schedule than everyone else.
Also this was in the late 80s/early 90s before cell phones and the internet.
My brother had a friend try this in university while studying for exams and, according to my brother, it basically made the guy lose his mind. Seems like a roll of the dice on whether it works for people or not.
There's a very active historical debate about whether a schedule like this, often called "biphasic sleep", was more common in pre-industrial societies. There's a historian called Roger Ekirch who thinks it was, starting in his 2004 book "At Day's Close - Night in Times Past". There's a bunch of criticism of it, since the sources are a bit ambiguous, or if he's generalizing from Medieval England (his main focus).
> I once changed my schedule for a month to sleeping 4 hours, waking for 4, sleeping 4, and then waking for 12.
That's pretty common for crew on yachts. I'd do that when racing or cruising ocean going sailboats. The crew gets split into 3rds, and everybody gets to do a 4 hour night watch. (For racing for me, that'd usually be only for a few days or a week tops. I twice spent ~6weeks on that schedule cruising the Great Barrier Reef.)
I learned typing on an IBM Selectric, my parents made me take a typing course before they would buy a computer. White-out or correction tape was how we fixed typos (or just didn't make them). If you were smart you wrote your words out longhand before typing them; you didn't "think" while typing.
The early computerized typewriters kinda sucked - you could do line level edits but the print quality was dot matrix or worse.
WordPerfect and WordStar were always pretty yuck IMO once newer generation products came along. I was pretty much a fan of Microsoft word even in the DOS days. (Even Multimate which was basically a DOS clone of a Wang product.)
I loved WP5.1 for DOS because of one feature: "show codes" - it made it trivial to understand why the formatting looked the way it did, and to fix formatting problems. Other than that it was not outstanding software :-)
I never used Word in the DOS days so I can't compare, but it was obvious that Word for Windows was written natively for Windows and it "felt" much more natural in the Windows of that time.
The NeXTstep version was _very_ nice --- looked and felt like a native app, but still had "Reveal Codes" --- nicest version of WordPerfect I every used.
Exactly. The big companies are scared of lawsuits and trying to get approval for something like that would be a nonstarter. As a matter of fact the device folks at the same company would be working hard to kill such an idea in its infancy because it’s already an uphill battle to sell always-listening or always-watching devices to consumers because of the creepiness factor.
And people also are terrible at math. Modern ML (regression & neural nets) are ridiculously good at predicting stuff you might be interested in, particularly when rich data sources like browsing and e-commerce histories are available; the decision to show the ad to you at some point almost certainly was made long before any audio-to-marketing pipeline could act on it.
Neither is a US company. TikTok is on their last appeal to the Supreme Court to avoid being banned in the US. Huawei is banned by the US government for many uses, e.g 5G infrastructure. Neither is a good example.
I think that the difference between that app and the proposed app in this branch of the comments, is that that app is just an LLM echo chamber for your journal. It would be different if you also saw posts from other people that you could react to, like on more traditional social platforms.
Maybe the right mix is to have an LLM-botfollower-army feature that you could purchase in existing social media apps, but your botfollowers are only visible to and only interact with you.
Also there should be a lot of different prompts for the botfollowers so that they don't all sound the same, e.g. the prompts would drive different personalities for the bots. Perhaps an algorithm could generate prompts based on archetypes mined from existing social media.
Nobody has the patience anymore to be presented to or read in any form other than bullet points and low information density charts.
I grew up before computers and learned to communicate in the absence of all the short attention span distractions that exist today. I remember the first time I picked up a Wired magazine and couldn’t tolerate the insane lack of continuity. I still cannot stand the video style of images projected for a fraction of a second one after the other.
But no one has the patience for my storytelling style. Congratulations if you got this far, most people gave up if they didn’t grok my point in the first two sentences.
Yes slideware is ugly and low information and boring and insulting to the audience, but some people, particularly in higher levels of management, just want to be spoon fed bullet lists and then feel like they’re making informed decisions.
In the age of LLMs there is no point in writing long prose. The more content is generated the more people will move to higher information density formats. Why bother with generating long text if the reader is going to summarise it with a LLM anyway? OTOH - citing Ben Affleck: "why would I bother to read something you didn't bother to write?"
> In the age of LLMs there is no point in writing long prose. The more content is generated ...
Wow, do you really equate long prose with "generated content"? Long prose is novels, deep non-fiction books, long letters, and much more. You can like them or not. In comparison "generated content" is sugar-coated garbage, like way too many social media posts. There was never any point in reading such "generated content".
I wasn't clear enough and I agree: art cannot be replaced by LLM (although this is heavy disputed by AI believers). Consuming art also precludes reading summaries generated by LLMs.
My comment was about "utility" texts (this is a context of this discussion, I suppose) - my prediction is that we are going to write shorter and more condensed texts to avoid overhead of LLMs use in generating and summarising text.
But why should there be two polar opposites: utility & art? Even when reading comments on HN I appreciate a a well written narrative with clarity and cohesion, instead of an assorted stack of bullet points.
The idea is that good writing actually makes you think clearer (both writer & reader); it’s not just a nice to have.
If that was the case people wouldn't use LLMs to generate long text only to use the same LLMs to generate short summaries of it.
What I am saying is not that good writing is useless - rather that good writing is _hard_ and people are lazy. There is way more bad writing than good in the world. Bad writing will be replaced by LLMs which does not make sense because it is still bad - and useless.
Good writing is going to stay but since it is hard it is (still) going to be rare.
In the end my hope is that bad (and useless) writing is going to be replaced by short, dense and useful format.
Of course - this begs the question: what constitutes good writing? Pretty good estimation is that a good writing is the one that is - generally speaking - as information dense as possible (ie. there is nothing you can take away from it without loosing some important information). And we are back to square one - it does not make sense to write anything longer than necessary :)
To add to that, text has more of "authorial intent" (debates on the demise of which notwithstanding) than other media.
Consider the visual rebus, for example, which is open to interpretation and depends on commonality of context in both producer and consumer, contrasted to a rigorous argument, which depends onoy on commonality of (technical) jargon.
Video ends up conveying information thanks to narration, while the visuals assuage boredom. Like an Adam Curtis documentary: it's essentially an essay read out, with clips and music overlaid to keep the audience from realising they're told, rather than shown, the argument.
Having the talking points as aides memoire on screen is nice in that it charts the course of the argument, but the map is not the territory, and we end up with significant information loss and knowledge gap.
I think that moving from the message in itself to its summation (i.e. from text to bullets) creates a knowledge divide between the producer (who knows more) to the consumer (who has access to less and can only divine the rest).
Its not that they want to be spoon fed, that is the ever decreasing bandwidth available as info and complexity in environment keeps exploding.
As the saying goes "glue people" just need to know a little bit more about coding than the sales guy, and a little more sales than the coder, little more accounting than the lawyer and a little more law than the accountant etc etc.
Could it be that such increasing specialization (sales, accounting, coder, glue…) is driven in part by decreasing attention spans and ability to focus? Maybe nobody is able to exert significant deliberate cognitive processing on anything beyond a narrow slice that they are already comfortable with.
It's natural. Rate of change is increasing in all fields. So when teams can't keep up, another team is setup to handle whatever is new and then automatically you get divergence in expertise, between the two groups over time. If you specialize in selling on amazon you learn totally different things, than selling in a store. If you specialize in mobile app development, a gap grows with teams who work on web or desktop. Info Asymmetry and Knowledge Gaps between specialists keeps growing with time. So the glue people naturally emerge.
The number of books, blogs, papers, textbooks, monograph, tutorials, reports, comments that are worthwhile reading will only ever increase though. Today is as information-lite as we're going to get. I feel like it's only natural to try and condense all this data (usually in a lossy way).
I hate PowerPoint but bullet points could be quite information dense. They lose effectiveness past 6 bullets due to readability reasons, but I still prefer them over fluffy low signal prose. I think at the end of the day, it all depends on the writer. I got to the end of your comment just fine. But I cannot and will not get to the end of a corporate jargon filled report or ppt.
> some people, particularly in higher levels of management, just want to be spoon fed bullet lists and then feel like they’re making informed decisions.
That caste likes to say, "I have people that do that for me."
I can actualy cope quite well with the slides that have 5 words on. But typically in the organisations that I have worked for, people will cram 200 words on a slide and then droningly reads them out.
No matter how often I explain that 'people can either listen to what you are saying, or read the slide - pick one' - it doesn't really sink in
A recent colleague often gave my team feedback like that. But we listened to him and all our presentations were much better afterwards. I hope I've internalized that idea somewhat.
reply