Is there any reason they are unpopular other than they don't have much momentum and it kind of sucks to type? I think they are cheap domains but have avoided them for the assumption that they just don't get the SEO of a .com
Not being judgmental. :) I've started a few sites and I always think about stuff like this, especially with e.g. IO domains being subject to reclamation by the countries that manage the TLDs. Seems like MOBI is a bit more secure being backed by Google + Samsung and a few others though
To those, like throwaway on this thread, who say " Give it 5 years and this "AI slop" will write better than your best trained journalist. Get over it."
How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
AI can hoover all that up and rewrite and expand, but surely there will be a need to have boots-on-the-ground?
If so, who pays?
I am already seeing AI giants using local media copy in output with $0 paid for the knowledge and information.
Those people have no self-reflection and in 5 years when their claims haven't come true they'll have already moved onto whatever the latest hyped up thing is. Just like how we still don't have mass driverless cars, and just like how Bitcoin didn't take over the world, just like how LLMs aren't going to take all our jobs. I bet there's a good alignment between people making outrageous LLM claims and people who shilled for those other things too, for how they were "right around the corner".
Ironically, an actually good use case of LLMs would be to scan peoples HN posts and compile a rating of how many hyped up topics they shilled for in the past, so that we know who to just kind of ignore when they're hyping up something new, as they don't have a good track record.
> I bet there's a good alignment between people making outrageous LLM claims and people who shilled for those other things too, for how they were "right around the corner".
I'm still waiting for NFT's to take over the video gaming space, its just around the corner!
Pppphhh we don’t make video games anymore. We just get the latest LLM to boil a lake so we can aimlessly wander an incoherent fever-dream! It’s the future of gaming!!1!
I remember a post on Blind where someone was passionately arguing that if we put house ownership on the blockchain, 2008 wouldn't have happened because a big part of it was not knowing who owned what homes.
Isn't that Microsoft's plan with Game Pass and buying the whole western game industry? You'll have a choice between Axie Infinity or Nikke: Goddess of Victory.
Non-fungible tokens are not even necessarily digital. They are merely tokens[0] that are not fungible[1], as the name suggests. Digital non-fungible tokens are stored in a digital database, of which blockchain is one kind.
Gaming is actually a pretty reasonable use case for NFT's, or blockchain generally. At it's core, an NFT is a digital asset with verifiable ownership. Many, many games have the concept of owning digital assets; the fact that they're not on a blockchain is an implementation detail. All the hype around monkey-ape pictures and microtransactions was offputting to a lot of gamers, making the whole thing come off like a big cash-grab (which it mostly was), but from a purely technological standpoint, blockchain might not be a terrible solution for games with transferable digital assets.
> the fact that they're not on a blockchain is an implementation detail.
For a good reason. The blockchain doesn't add much over a database. Players being able to sell their items? Steam does that without a blockchain, albeit not for real money. Being able to take items between games? Almost no devs (or players) really want that. But blockchain would add difficulty changing things if there's a bug with the items (or not, but you lose the "security" it's supposed to add).
which has rather good trading mechanics in the game (for non-NFT objects) and has a total of 20 or so NFT objects in the game. A handful of players own most of them, for instance there is only one type of watch you can wear in the game and there are only five of those and four of those belong to the same person. My son did see someone wearing a watch.
You can't say the NFT ruined the game for other players but you also can't say it is a success, I wouldn't say that until you have broad participation in ownership and trading.
If you invested hundreds of hours into a game and accumulated a significant pile of rare items with in-game utility, and then decided you were done with the game, would you rather a) have the option to sell them for real money or b) have the option to sell them for in-game currency (which you will never use) or c) not have any marketplace options?
You're talking about a closed system for a single game though, what exactly does a blockchain actually give you in that example?
All 3 of those are possible with a traditional database.
Even if they wanted to provide a system for sharing those assets and other marketplaces being able to built on top of it, a blockchain still doesn't actually provide any benefits.
A lot of them intentionally are not supposed to be sold for real value because in many jurisdictions, combining that with lootbox or drop mechanics would immediately make the game subject to gambling regulations.
Game dev companies don’t want that, they want to keep selling you stuff. See: games like the Sims
A real-world market attached to a game would be the fastest way imaginable to suck all the fun out of a game.
I don't want to play games where 90% of the players are using the game to try and beat capitalism somehow. There are 100 reasons why this'd destroy gameplay.
* The majority of players would be concentrated on the highest value grind tasks.
* Scams would be rampant. They are already rampant in games without this mechanic. Now imagine how bad they'd be if your items had actual monetary value hooked up to them.
* Gang like behavior is pretty much guaranteed to emerge. So, it wouldn't be just one player grinding like crazy, but groups of players dedicated to making the game miserable for anyone grinding on their turf.
* It would encourage actual real-world problems/crime. Now instead of being able to meet up with people and talk about the game, you run the risk of being mugged or stalked for game currency.
Just a really terrible idea.
And these things I'm talking about aren't imaginary. There're already more than a few articles about games like EVE Online and WoW where the marketplace of those games has led to the problems I listed.
EVE's "Real life threats" policy didn't show up because it wasn't a problem [1]
I'd rather not have the marketplace, honestly, and I'm someone who got more than I paid from TF2. Trading items for items isn't so bad (maybe), but we don't need to make things even more pay-to-win.
Games should be about having fun, not second jobs, unless you're playing a simulator where your fun is doing the job.
The discussion was more about in-game items than game licenses, but that falls under "almost no devs really want that". Devs want to be able to ban problem players from their games. Devs want (or at least would prefer) you to have to buy a new copy of a game instead of a cheap used one. It would be nice if we could transfer the licenses, but the developers don't want that.
And it also doesn't require blockchain. Valve not allowing it isn't a technological limitation, it's a policy. The system can't stop you from leaving a sheet of paper with your username and password on it, but Steam can ban you if they somehow find out.
And how do you enforce that Valve implements and respects the blockchain? You can't, and if you do you can just force them to respect a traditional database.
Depending on the use case, you don't need to store the entire implementation of the item on chain, just a reference to it. All the implementation could still be stored game-side. You could also apply nerfs/buffs over the top of an item at runtime.
> What if they release an in-game item that turns out to be completely over-powered and want to change or remove it?
I'm not sure how/if real "crypto games" do it, but it wouldn't be impossible to have a list of "invalid" items the smart contract references that can't be used unless you use a second contract the developer set up to exchange them for the nerfed version. And I guess that could enable community-created formats with their own banlists (if the game is actually decentralized enough for that to be possible).
But then we've had to do a lot of extra work for a messy version of what we can already do with a normal database!
From a purely technological standpoint, a boring database does everything an NFT could for transferable digital assets.
This is the classic "solution looking for a problem" you see so often with NFTs and blockchain generally. The proposed problems (in this case verifiable and transferable ownership) already have mature, satisfactory solutions.
NFTs don't benefit consumers (because the tangible value exists in a centralized server somewhere anyway) and they don't benefit publishers (who can already implement those features if they cared to). They benefit people that want to get rich speculating on NFTs.
So, the use case I'm thinking of is more along the lines of non-cosmetic assets. Take FIFA Ultimate Team, for example. The way it works currently is that each player in the game has one or more "cards", or sets of stats. Special versions come out all the time which have significant impact on gameplay.
FIFA has an in-game market where you can buy/sell/trade cards, for in-game scrip. Want to sell your special Killian Mbappe card you just pulled? Congrats, you can get 1,000,000 FifaCoins, not redeemable for real money. Of course, you can certainly buy FifaCoins for real money.
A blockchain implementation could potentially allow people to sell that card directly to somebody else, for real money (or more specifically, crypto tokens that can be redeemed for real money). Would you rather have 1,000,000 FifaCoins, or $100?
Whether or not this is desirable to games is a different question, but the game is already essentially pay-to-win, so it's not like it would ruin the game from that perspective. It already has microtransactions. The main reason you probably won't be seeing this is because it would cut the publisher (EA) out of the loop, which is very, very lucrative for them.
The best example of this is probably the Diablo 3 real money auction house. It was a launch feature of the game that would allow items to be sold for real money or in game gold.
It was removed about eighteen months in, because it made the metagame significantly less fun. Switching from a localized rarity table with some light trading to an efficient global market meant that finding anything less than best in class equipment (bounded to level) was a disappointment compared to what could be bought on the market, and had to be to avoid flooding market supply.
Getting rid of the market and implementing more generous local loot algorithms probably gave the game at least 5 years more life.
if EA wanted to set it up so you could sell them for real money, with ea taking a cut, I think they could. of course they don't really want to let you do that, but even putting aside banking steam let's you sell cards for steam wallet money (which goes back to valve) and let's you buy cheap games for "free".
it's never really seemed like a limitation that needs a technical back end to fix it, it needs people to want to set it up.
Why would any game publisher relinquish the insanely high profit margin DLC/skins they sell to an outside third party? It makes zero sense, they can just use a SQL database and keep all of the profits.
In short, publishers aren’t going to give the crazy margins they make on digital in-game items to a third party NFT blockchain, they’d be insane to do so.
There is no use case for NFTs at all. The "verifiable ownership" is pointless when you could just use a traditional database, as every individual game has to respect the blockchain - they could just respect the database instead, and not use the power consumption of Belgium to do it.
> Those people have no self-reflection and in 5 years when their claims haven't come true they'll have already moved onto whatever the latest hyped up thing is.
Yeap. Just like there are interesting use cases for blockchains (note that I said interesting, not necessarily practical) and machine learning, there are interesting use cases for LLMs.
The problem is that people then come up with new applications "really soon now" and use that to hype up wallstreet.
It's easy to say ${new_thing} is going to take off (takes a while to be disproven), it's even easier to say ${new_thing} _isn't_ going to take off (takes a while to be disproven and most things don't take off), but it's hardest to actually have a clue when you're talking about something which changes life forever (like the internet), a curiosity [for now] (like Google Glass), or just a scam from the ground up (pick a scammy cryptocoin) immediately when you see it.
The Internet is "just" the third Network. Both its predecessors (the Universal Postal Union and the Public Switched Telephone Network) are still hugely important, it's amazing that people don't understand the significance until it's shoved in their faces as a result of Tim's mediocre hypermedia software (the "World Wide Web") and even there are plenty of nay-sayers for years.
The Network is a big deal, as big a deal as say, pornography or literacy. But while iterations of the Network are significant they're also sort of predictable. A society which builds the PSTN is going to build something akin to the Internet next.
I agree both are easy to say, but IMO ${new_thing}s fail like 80% of time and, even it succeeds, take much longer time to find its useful usages and it's much narrower than we expected at first. So I think it's better to have skeptical or pessimistic attitudes toward ${new_thing}s by default. BTW I'm impressed by LLMs for some tasks and it makes my job easier, but still feel it's useless for majority of my daily tasks.
> it's even easier to say ${new_thing} _isn't_ going to take off
Not around here it isn't.
> but it's hardest to actually have a clue when you're talking about something which changes life forever
Most of us saw the Internet coming from a decade away, Google Glass was never going to be a thing, and once cryptocurrency "mixers" got invented everyone saw the writing on the wall.
> Most of us saw the Internet coming from a decade away
If that was true as we'd like to believe for most of us HN would be more of a retirement community than a startup community. You can blend a bit of how everyone knew which parts of the internet were scams and how that wasn't all there was to it in that as well.
Everything seems obvious in retrospect, down to how "obviously it was only going to explode larger after the dot com boom" or "getting into ads would be the way the most money was made". All of the areas that failed or were different than expected get forgotten. And most importantly...
> Google Glass was never going to be a thing
Given many years a lot of things end up working out in ways that weren't really expected but are, retrospectively, "obvious from a decade away". Much of the time this doesn't even require new technology of the day, though it may be better with the advances. Will that be a Google Glass style thing? I doubt it personally... but I can also manage to temper that with my certainty not being absolute.
> If that was true as we'd like to believe for most of us HN would be more of a retirement community than a startup community.
There’s a difference between saying “something like this will exist” (internet, cryptocurrency) and “this particular implementation will win the market” (www stock picks, coins). People here saying crypto isn’t going to take off (as a currency) mean the whole concept, not just one particular implementation.
Do you have anything other than a single counter example?
> for most of us HN would be more of a retirement community than a startup community.
What data have you seen that makes you believe it isn't? What makes you think it's a "startup community?" You're also forgetting just how long it takes for new technology to penetrate and how often it gets _integrated_ with what already existed. Merely seeing that this would occur did not magically create new opportunities. Not all technologies are transformative as you suggest.
I mean it was obvious that people would eventually order pizza using the Internet. That didn't mean you could run out and start an Internet enabled pizza business and then cash out.
> but I can also manage to temper that with my certainty not being absolute.
Battery life, weight, and connectivity were _obvious_ issues. If you think like an engineer rather than a temporarily embarrassed venture capitalist you might find that confidence very easily.
> I bet there's a good alignment between people making outrageous LLM claims and people who shilled for those other things too, for how they were "right around the corner".
Have you ever played the "scroll back far enough" game on e.g. Twitter?
So many guys (it is always guys) with .ai in their posts used to have .eth in their posts. They used to be all-in on NFTs.
Grifters gonna grift. Salesmen are much more likely to believe sales pitches, etc.
I would also add that "Get over it!" is a reliable tell for an awful person. It means "I do not wish to be confronted with even the smallest, most unarguable negative consequences of the position I have enthusiastically bought into".
We have driverless cars, Bitcoin just hit $100k, and LLMs are undoubtedly taking over jobs. Hyping something that never materializes is one thing, but these are not that.
I will admit, there were a lot of people saying ridiculous things about crypto in 2021. Many of those people were very wrong, but still rewarded with bitcoin hitting 100k.
Those same people say it will hit 200k inevitably. I think all their reasoning is wrong, but yet they have been right. Curious what HN's opinons are.
> I think all their reasoning is wrong, but yet they have been right. Curious what HN's opinons are.
There's no way to predict the price. Everyone who's claiming to make a prediction is just guessing.
There are a number of influential people who are heavily invested in crypto, and so long as they have the ability to keep money funneling other people's money into it, prices will continue to increase. So I think the safest guess to make is that it prices will continue to rise for a least the next few years.
I kinda feel Bitcoin hitting $100k does not mean its success. I think it proves to fail to become a currency as intended at first, and instead it becomes a gambling target and political tools for billionaires. That's far from the original purpose.
> Just like how we still don't have mass driverless cars, and just like how Bitcoin didn't take over the world, just like how LLMs aren't going to take all our jobs.
In San Francisco, we do have mass driverless cars. It's hard to drive across SF without seeing one.
Bitcoin didn't take over the world, but it's well on the way to taking over the administration and Congress through sheer lobbying spending.
LLMs aren't taking over all the jobs, but if everything you do for money goes in and out over a wire, less people like you are needed.
> Just like how we still don't have mass driverless cars
Mass? No. Not yet.
However, Waymo is doing a pretty good job at this point. I'm no longer "It's indefinitely 15 years in the future" about self-driving cars.
My interactions with Waymo cars leads me to believe that they are probably better than the "average" driver, nowadays. They almost certainly have passed the "average teenager" bar long ago.
> How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
I noticed that most of the journalism that I care about ~drifted~ from the writing to non-structured media (e.g. podcasts, videos), where in writting we have only things oriented to the SEO and future reference and historical purposes, and the actual red meat content comes in "conversations" or in aired media in youtube channels.
Back in the day, a journalist could sit in a hearing or in an audience, and you could read at the weekend a deep dive with key points in the trial, points made by the prosecution and defense, the verdict, and the entire trial dynamic and buildup. Currently you have in writing only some high-level summary with no specifics and maybe some journalist opinion, and the entire breakdown you can have it on Youtube.
For me, this is one of the two saddest trends that I see in modern journalism (another one is the lack of investigative journalism revealing political corruption), where you can only get in the depths by going to the prolix audio and video media formats.
At the local level, there is (almost) no modern journalism. Large media companies have swooped in and gobbled up pretty much every major newspaper, local tv news, local radio news. They've then fired the journalists and switched everything to sites that push mainly national stories mixed in with 100 clickbait marketing garbage ads.
And for national journalism, it's 90% about access journalism. Every interview is softballs, no follow up, etc. Journalists almost never press for an answer. And it's just so absurdly lazy. Rather than actually reporting or investigating anything journalists just republish press briefings.
> Large media companies have swooped in and gobbled up pretty much every major newspaper, local tv news, and local radio news.
I think the main issue is related to the funding of those projects. As a non-American but former resident of San Francisco, I would like to see more things like [1] (regardless of the political side) for local politics.
It's a pity that the national news is pushed so hard that local councils and mayors can do almost everything without any pushback, even being the kind of politics that has a direct and immediate effect on most normal people.
One thing in tech I find interesting is I run into a lot of people that view everything as a tech problem. The AI can be a local journalist if it gets good enough and on one hand that's always true, sufficiently good tech can do anything, but it seems weird to have it be the only suggested solution by some people I know.
It's especially confusing to me in situations like this where the tech is currently exacerbating the problem and the underlying reasons why aren't going to go away just if the tech gets better. On the current trajectory I don't get why the technology getting better will suddenly have the AI do substantive local journalism instead of just higher quality garbage.
In conjunction with actual journalists I could see it being useful but the article seems to be talking about a site with fake bylines so that doesn't seem to be the case here.
What’s the practical alternative? Viewing the problems through a socioeconomic or political lens just makes them seem intractable. At least techies can influence technology, either through entrepreneurship or open source contributions.
As myopic as it appears, the approach at least scales well and has drastically transformed the world in the last few decades, for better or worse. The problem is that we hype up technology as a panacea when it’s at most an enabling component for improving human effort. At least we can do something to effect change if minor.
I think it mostly just seems intractable because people at the end of the day don't care about local journalism (at least not enough to pay reasonable amounts of money for it). At the end of the day I think there are some problems that are pretty simple with collective action but people just don't want it badly enough to actually sacrifice anything (in this case, money) for it.
Where I grew up had a local paper, and the community was more than wealthy enough money to pay for it but circulation was ever decreasing. In my naive view it died a death of apathy more than anything else and it's not surprising there's not a satisfying solution to it.
I think it's just another example of enshitification.
Newspapers can't make the bills with just subscriptions, so they start selling ads. Ad revenue starts dropping so they add more ads. More ads cause less readers. Cost cutting measures erodes at the quality of journalism. And then a big media company swoops in, buys the company, fires basically everyone, and turns the newspaper into one of the uniform newspapers that they also own eschewing all local issues. That further drops subscribers and trust which simply leads to more misleading garbage.
Throughout the process, yes people cared less. But also quality decreased which made people care even less.
My parents who have had a newspaper subscription for 50 years have cancelled their subscriptions a couple of years ago because of how garbage the paper is.
> How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
It can't. And the notion that it will be able to "in five years" is completely without any basis. It's possible, I suppose, but anyone who claims this with great certainty is not a serious person.
Demand disappeared for good local reporting, long before any LLM. Even NYT & WSJ struggle with their subscriptions, and they have a huge addressable market.
The "slop" is just a reflection of this bottom feeding attempt to get some tiny bit of ad revenue. Using an LLM or a "pay per article" contracting fleet looks very similar.
Most local publications don't pay many if any full time journalists either way. I struggle to find great journalism on any local level around me - and discovery is difficult when anything reasonably good is usually a substack behind a paywall. But even then, they aren't pushing on local issues in person. None are going to the Mayor's press conference and directly asking them questions.
> How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
Is that a thing with local journalism now? Actual journalists still doing stuff like that? Lets not pretend that a lot of good writing or journalism is involved with local news papers. It's not a particularly hard job for an LLM to do better.
In any case to answer your questions:
- yes you can let an AI listen in on meetings and conversations and they'll do a decent enough job of transcribing what was said and summarizing key points.
- you can also use them to prepare for interviews, get some background information on people, companies, etc. and prepare some questions. Perplexity is great for background research like that.
- and if you provide all of that as context, it will do a half decent job of writing an article. You might want to do some prompt engineering to tune style and content and maybe fact check a thing or two. For best results, maybe use one model to generate and another to fact check.
You are probably already reading a lot of stuff that is AI generated; or at least AI assisted without realizing it. It's not always that obvious. The future is last year. Already happened. What you are seeing is the more obvious/lazy stuff. Which is of course a thing because good enough is good enough and the benchmark for that is very low with most media corporations. And being factual isn't necessary something they value either.
AI may not replace good journalism but it will flood the world with garbage so that it makes good journalism harder to spot for the average individual.
There's an incredible amount of local journalism--especially outside of major cities--that was already incredibly poorly covered. There used to be a labor of love type newspaper in my town until the publisher got ill. At this point, there's basically very little way to get information on the various town meetings outside of major issues and mostly not even then. I'm not spending the time and effort to attend a lot of random board meetings.
There are always two sides to news -- there's the consumer-facing text/video/whatever, which AI can create. But there's also newsgathering -- the reporting, which requires shoeleather, phone calls, building relationships, etc. AI cannot do that.
(Disclaimer - building a news platform featuring human journalists' reporting - https://www.forth.news)
the "carpenter media group" is paying. theyre a conglomerate from the southeast that started buying up small town news agencies across the country in 2023, and began laying off 50% of staff to replace them with AI. please see my other comment in this post.
- the family trust of 22 year olds in Columbia Journalism School. Don’t take this too literally. Journalism school is a metonym for paying your dues.
- the husbands and wives of the older journalists
- millions of Boomers paying just four publications that are diversifying into media companies with video games and recipe lists, so not really journalism alone
Should journalism be philanthropic? It’s an interesting question. I’m not worried about AI writers as long as parents keep paying their kids’ rent to break into writing, and as long as New York and Los Angeles remain attractive places for young people to live. My point is that there is a status quo where journalism is already essentially a charity, so anything is possible, it is totally incorrect to characterize journalism as profit motivated, because if it were it would have been gone long ago, and it isn’t for lack of free labor, because most journalism is done for free.
> How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
How is any of that relevant to getting people to read your articles? Seems like getting rid of all those will make reporting a lot cheaper. That way, anyone can start a newspaper and report whatever they want!
I'd argue that you have a much better chance of a consistent reader-base if the contents of your articles have no base in reality.
> How does AI attend meetings, ask questions of local officials or bodies, provide in-person reports and original fact based copy?
Streams? The AI summaries of YT videos are already pretty good. And asking questions is probably easy enough as well, assuming the meeting is using a platform that allows for questions to be asked. Text-to-speech can be used on platforms like Zoom or even a simple phone call.
An AI can go through all the breaking footage posted on social media. An AI can join a phone call with 400 individuals and retain complete notes on everything that is happening. An AI can go through every single expense report, every month for a decade to find the one thing that explodes into a story.
An AI journalist won't be a 1:1 copy of the human, but it can definitely do actual reporting.
How useful will that AI be if most breaking footage on social media will be generated by other AI? When the phone call is joined by 400 AI assistants instead of humans (assuming such phone calls exist at all)? When all those expense reports are also AI generated?
The main problem in the next 5 years and beyond will be filtering factual information from AI generated garbage, which is a difficult task for humans in traditional journalism. Until we solve that problem, more AI content is just adding more noise.
I don't doubt that the scale will improve but I am very skeptical that the reasoning will see the same kind of growth, and at scale, in the example you're describing could look like a shitshow. It's possible the reasoning will improve but with a paradigm shift which I don't see it happening at the moment.
I think the point is that it might not happen in five years but it will certainly happen in 20, and we should still be as concerned about it then as we are now, because such a powerful technology could definitely spin out of control.
If you live in a town with a major industry that industry tends to get a lot of free PR from local news who eat up all their releases at face value. AI writing would surely do the same but globally.
I'm sympathetic to the way you're looking at this, but imho the end state will be similar to most other automating technologies.
The simplest, lowest-effort work will be automated (sitting and transcribing), and humans will reapply themselves to (more productively) do higher-level tasks.
I'm sure a local beat reporter would be thrilled if they could get notes from all the meetings yesterday.
And asking questions? Will the AI have hallways conversations, historical context and relationships with previous office-holders that inform the questions?
A reporter is supposed to care about accuracy, an AI only cares about generating words that make sense and fit the page. I can see why the owners align with the AI more than the reporter.
We're living in an unfortunate time when people have been trained so thoroughly to hate institutions that they've lost basic understanding of how those institutions work. And how they still serve us, despite their flaws. The institutions have become nothing more than abstractions that can be rhetorically written off and replaced by either outsiders from the establishment, or in this case algorithms.
don't forget the people who try to exploit the institution for their own gain. Thus ensuring institutions are bogged down in bureaucracy to prevent that.
I'm not sure SOTA is at the most insightful questions yet, but those are rarely the ones that get answered at briefings anyway.
Point being that it's not "human-or" but "human-and".
A local reporter who spends their night reading summaries of the day's meetings, diving into the raw recordings where needed, and then writing prompts for tomorrow's agents doesn't sound like a terrible future.
As the quip about local journalism goes: a lot of it is sitting around at meeting #324, to pick up the one interesting tidbit.
With local news having been apple-cored revenue-wise by social media, that's now generally just not being done outside of major metros.
So I'd welcome anything that promotes more transparency and local reporting!
Transcribing is almost at the end of the journalistic process. The hard part is getting the facts (correctly), checking and cross-checking, calling sources, trying to get back channels, in many cases have boots on the ground, and then summarize all those things and put them all together before editorialization.
However, the "5 years" timeline which yeah its crap with what they are saying. I do worry that what they are actually hoping (not paying for writers) for will actually happen. Since there is a lot of money in not paying people and pushing content that keeps people reading.
Just regurgitating information over and over again in different ways. Or taking stray quotes or whatever as the information.
I worry that the question isn't if AI is going to get better or how its going to answer some of those questions that you have when the people running the sites for news clearly already don't care and its going to happen anyways. It doesn't matter if the technology is any good or it's valuable.
Maybe at some point people will wake up and realize that the regurgitated content serves no real value, but at that point how much have we dug ourselves into this hole already.
I know that some people compare this to crypto but there is a much different acceptance level and... ease of use here that is making it actually gain traction.
> AI can hoover all that up and rewrite and expand, but surely there will be a need to have boots-on-the-ground?
You missed the funeral for beat reporting by about 10-15 years. Boots on the ground has been dead since people stopped buying newspapers and blogging took over everything. This is just a further cost reduction.
Most “journalism” these days is done from a desk. The Migrants eating Cats story for example — the national networks ran statements from certain local officials but didn’t actually go there and interview or investigate directly. They relied upon statements from public officials to refute the story. (Not saying it was a true or false story, but the major “news” networks certainly didn’t spend much, if any, time there — they didn’t investigate anything; they reported what public officials said.)
Many stories might as well be AI: feed into it something Trump said, then the algorithm creates a story with lots of “without evidence” qualifiers and then call it a day.
The media rarely investigates anything — they report on what others investigate or choose not to investigate and that’s printed as fact depending on who made the claim.
> The Migrants eating Cats story for example — the national networks ran statements from certain local officials but didn’t actually go there and interview or investigate directly.
This is just trivially debunkable. Video evidence to the contrary - with CNN reporters wandering through town, interviewing locals - takes seconds to Google.
> The woman behind an early Facebook post spreading a harmful and baseless claim about Haitian immigrants eating local pets that helped thrust a small Ohio city into the national spotlight says she had no firsthand knowledge of any such incident and is now filled with regret and fear as a result of the ensuing fallout.
For a story like that how do you know "no one" went there? Do you read every newspaper and watch every news report?
It's been debunked to death and was a rumor that was posted to Facebook and then parroted by right-wingers looking to gain votes (it worked). It's a theater of the absurd that we're talking about journalists not doing their job and not the president that either lied about it, or fell for it and amplified it.
These problems are trivial when considering that we're already living in a sci-fi timeline where this technology was largely considered impossible 10 years ago. You're asking how it's going to know to ask the right questions, and to whom, when getting a system to ask questions in a general sense was the hard part. If you can't imagine humanity overcoming this hurdle, you lack in imagination.
There isn't a hurdle here. These problems are fundamental to the technology.
If you told me 20 years ago that you could, by connecting a nuclear power plant to a virtually infinite number of GPUs, produce a machine capable of predicting the next token in a sequence, I'dve believed you, but also would have said that it would be an astoundingly huge waste of resources to use to get to that result.
People seem to ascribe some magic here that isn't happening, because they're surprised by the results.
> People seem to ascribe some magic here that isn't happening, because they're surprised by the results.
I believe this is the result of drinking the kool-aid. At this point though, I'm really confused on how AI has any jobs left to do because blockchain was going to solve all the problems.
I don't think it's a useful way to frame it. Model training is very compute-intensive. Generation isn't. You can run it on locally on consumer hardware. It's just not very monetizable, so we're converging on a black-box "in the cloud" approach.
If we're focusing on energy consumption, using an LLM to generate a newspaper article probably uses less energy than a living journalist would use. The morning commute alone is probably "worth" a dozen articles.
The problem is different. First, that journalist is still alive and consuming resources, just out of a job. Second, because you now have a very cheap way to generate an infinite number of articles, and there are commercial incentives to do so, the "dead internet theory" has a good chance of coming true.
A major failing in futurism is not accounting for both quality and cost scaling, especially supra-linear.
If you'd told someone with one of the first automobiles that they'd eventually be massively personally-owned, have access the majority of countries via paved road networks, and be refillable along the way, they'd have laughed at you.
And yet, that's what we've all lived in.
It happened because automobiles were massively useful.
It's difficult to imagine a scenario where basic-level cognition doesn't also scale, because a lot of the mental tasks we do every day are dumb and low-value.
Nearly every example of futurism I’ve seen in my 40+ years of looking at future technology has been people over accounting for scaling rather than ignoring it.
For example they see a prototype and then assume in 10 years everyone will have one.
Or they look at a proof of concept and don’t realise that that last 10% of refinement actually takes 90% of the effort.
What people usually miss is that scaling is the problem, not the solution. Prototypes are easy in comparison.
I am not joking when I say that the autocomplete on the Microsoft HD Zune in 2010 is the single best autocorrect implementation that I have EVER experienced. You could half pay attention to a conversation, hit the right QUARTER of the keyboard with your giant thumb, and it would ALWAYS guess right. It was literally magic, in a way that no technology I have experienced has compared to.
Meanwhile, my modern "AI" based Google keyboard autocorrect can't even handle the average case where only the first letter of the word was a mistake. It will only suggest corrections (and then auto-choose those corrections based on rules that change monthly it seems) that use the same first letter that I accidentally hit.
Also, my mother had a dirt cheap LG flip-phone in 1998 that, after 10 minutes of training, could dial any number and even dial any of your contacts by name with your voice.
In the mid-2000s, Microsoft had a BUILT IN function to Windows XP for dictation. You can see it do poorly in a demo, which is funny, but in actual use, without any cheating or prompting or massive compute resources, it correctly understood you 80% of the time. The state of the art in the industry could do much better than that without "AI" at the time, and cost a pretty penny for people with disabilities mostly. However, that dictation API is STILL IN WINDOWS, Microsoft just kinda hides it because they want you to use the Azure based one. It is super simple, with a powerful yet understandable API, available to any application that runs on Windows (like, say, games) and completely free to use. More importantly, if you have ANY knowledge about the expected use case for your app, you can feed the dictation engine an easy to generate "grammar" about the structure of commands, and it's accuracy can jump up to near perfection. When I played with it, I was able to build a voice controlled app with a single page of code and like an afternoon that easily turned my voice into text. It was literally harder to control the legacy Win32 APIs to inject input into queues for my use case: Voice controlled game input.
Amen! Thank for that encouragement. I think you are right. I need to be calculated and disciplined about it. A short-break would do me some good, and I'm sure my family would agree :)
I don't think I've ever seen a page that large, outside of streaming videos and such. What an interesting time, especially because this could have been avoided by downscaling two images to fit the actual plausible display sizes.
I was on a site owned by Lyft the other day and I couldn't understand why the page was taking so long to load, I thought my connection was fried. Then I looked at the size of the images. All the images are 20MB+ each, even the thumbnails:
+1. I’ll never forget when I had an account with them, set up FIDO 2FA, and lost my password + email access for the account. I emailed customer support asking if there was something that could be done, and if not, I was fine making a new account (I had no domains at the time, and if I had, payment method confirmation would’ve probably let me get it back). Within days, their engineers added that feature, tested it, and rolled it out…
...and have published a 'disheartening' invoice dated 31st Oct 2022 - then work with the company for another ten months. Apparently content to be on the "dirty side of the Internet" and doing something that made them think "Something didn't seem right."
reply