People expect a binary response, is it AI generated yes or no. But it's more complicated than that. For example, if you see an emdash, it's probably AI generated. But it can also mean the author used it for fixing grammar or tenses. LLMs can't help but try to help. The same for it's not X, but Y. Sure it's a known pattern, but it's not like people don't use this trope all the time.
In my experience, Pangram is great for detecting an author who is trying to pass someone else's work as theirs, or if they are tackling a subject they have little to no knowledge in.
"We own the glass" that's what really stood out to me. I can still disconnect my TV from wifi. But what scares me is when every device comes with their own esim and cellular radio.
When you buy a BIC pen, you own all content created with that pen, not BIC. This needs to be applied to all other tools; computers, TVs, AI agents, ...
With out the user the content would have never been created. Users are the content creates and now the push to remove their owners ship of their actions and move it to the wealthy.
You shouldn't be really scared until all devices follow the pattern of the iPhone (and maybe new flagship Android phones too, I'm not in that camp).
New iPhones are expensive, more than $1000. Yet most people who get them finance it through their cellular carrier, usually 2 year plans, so it's just $30-$50 tacked on to the monthly cellphone bill. So ... no one cares unless it's going to double their bill.
Let's say the RAMopocalypse and other AI-driven chip pressues drives everything up even more; let's say now-cheap TVs are now $2000 or $3000. No biggie. Your cellphone company can offer you an eSIM-locked TV with bundled network and services, on a 5 year plan with free lifetime service and replacement, adding just $30-$40 to your bill for 5 years. Maybe it will include something like Family Tracker Life360 Intercom as a value-add - talk to your children to/from the TV to/from any phone on your plan, anywhere, anytime. Safety, etc.
Same thing with your laptop, home IoT/security devices, etc.
If ATSC 3.0 dies and gives way to 5g Broadcast, it'll be another enabler for something like this.
> But what scares me is when every device comes with their own esim and cellular radio.
Cars already do, and the solution is well-known: find the bug, and squash it. Yes, your car will still work, and you cannot be denied warranty because of this. The cellular modem is more or less accessible depending on the make and model.
Then they'll lobby for the eSIM to become a legally mandated component and outlaw use of a product without one. And lawmakers will be happy to oblige as long as they get to use the eSIM for their own purposes.
Correct me if I'm wrong, but I think the quote was taken out of context when they were speaking about vertical integration. "We own the glass" was meant to say that they manufacture the panels themselves and not source them from a 3rd party.
How do you even measure GEO, sometimes I try searching for my app and sometimes claude recommends my competitor that has 10x less reviews and is objectively worse saying is the most popular.
Because Google search isn't a hypertext document, it's a web app that presents dynamic ephemeral content to the user. There's no need for stable hyperlinks, since either you will click them within a few seconds, or they will disappear forever, and it shouldn't be a problem to indirect via Google when clicking them, since you just made a request to Google to get the link in the first place.
I like the trunk analogy, the branches can only grow so much before the tip the entire tree over.
I usually use the example of a blue sky. Write that the sky is blue. That's it. Simple yet the temptations to paint the sky with clouds, trees, and emotions is hard to resist. Always come back to the writing, and endure that you clearly wrote that the sky is just blue.
This is really fascinating. Even the AI companies have incentives to reject AI generated content. It's like they want you to use AI for everything, but they don't want AI output fed back to them.
At work, right after an AI training, we were asked to use our "authentic" voice when writing mid year reviews.
Turbos don't inject exhaust air into an intake. Exhaust air spins the hot side of the turbo which causes the cool side to also spin. The cool side sucks in fresh air. Not exhaust air.
Your comment would make more sense if you mentioned exhaust gas recirculation (EGR) system.
Or maybe it's good for everybody to use LLMs for some things but not other things, as opposed to some people should use it for everything and others for nothing. These companies are definitely using LLMs for coding.
The law is too slow. It's like a horse carriage in the age of twitter. Why can't they expedite for special cases? Not even defending Antropic or any company. Just that if a tweet can cause damage in seconds, the law shouldn't be too far behind.
To a certain extent this exists, but given that SCOTUS has been stacked with GOPers…
> The shadow docket is a break from ordinary procedure. Such cases receive very limited briefings and are typically decided a week or less after an application is filed. The process generally results in short, unsigned rulings. In contrast, merits cases take months, include extensive briefing and public oral argument, and result in lengthy written opinions detailing the reasoning of the majority and of any concurring and dissenting justices.
It's not unusual that the "shadow docket" orders have limited briefings when the extent of the ruling/order is usually procedural, as opposed to ruling on the merits of the case.
It should be noted any justice can write a dissent in a "shadow docket" ruling. If a justice finds a ruling incorrect, nothing hides their criticism of it from the light.
The problem is that the rationale is hidden, not that criticism of it is hidden. If the Supreme Court doesn't explain why they make momentous, weighty decisions, that leaves the rest of us with only speculation, both in understanding how the decision should be applied to other cases and evaluating whether the decision made any sense at all.
(I agree that there's nothing fundamentally wrong with the even more recent trend of shadow docket decisions that do come with opinions attached, and perhaps law students in the 2030s will learn about the "expedited docket" as a normal way to handle cases.)
This mechanism means that regardless of how awful the Trump admin proposal one SCOTUS justice can "temporarily" unblock it giving no reason at all, and then the "real" merits case can be slow-walked.
Slow-walked, conveniently, until the admin has finished doing whatever the original lawsuit claimed was illegal, and then the government argues it is impossible to undo after the fact.
Absolutely nothing about how the Supreme Court works nowadays has respect for standard process or decorum. It's merely a shield to hide behind when it's convenient, to be ignored when it's useful.
I was trying to avoid editorializing by just stating facts that demonstrate the process is beyond some issue of reigning in lower courts and is instead leveraged by politicians to achieve their goals.
I can also editorialize.
In my view, Trump is leveraging the shadow docket to get decisions over turned by members of his own party who don't need to provide detailed (or any) explanations as to why or information about who voted for what.
It indicates one of two things: either the orange fuhrer is being sabotaged and having to defend himself, or that he's doing blatantly illegal things and even a court stacked in his favour by him is trying to reign him in.
SCOTUS is not stacked with justices who are just GOPers. There are lots of Republicans who absolutely value the rule of law in this country!
The SCOTUS is stacked with reactionaries, people that want to take us back in history. They quite literally undid 60 years of civil rights legislation!
This is not a Democrat vs. Republican thing, this is is a very small minority that has taken control of an entire branch of government and is abusing their power.
Mostly, the law is not creating the right incentives. In this case in particular, I don't think it's so much a matter of speed: I suspect the current US admin would still sow the chaos they do to the same degree if it was ruled illegal within a week instead of within months.
You're right: the current administration would find a way because they don't care what the law says. The law, to them, is a system to be hacked when necessary.
Many Americans take for granted the idea that the US is a "nation of laws" and that "rule of law" governs above all else. But the reality is this: there is no such thing as a perfect legal system.
You can have a strong legal system by design (good oversight, various checks and balances, etc.) but in practice, any system is only as good as the most powerful people are decent and respectful of it.
If you have someone in the most powerful position (the executive) who doesn't respect the law and who isn't willing to, with wisdom in his or her heart, exercise restraint with the understanding that protecting the integrity of the system is ultimately more important than getting your way, the law stands no chance.
>Many Americans take for granted the idea that the US is a "nation of laws" and that "rule of law" governs above all else. But the reality is this: there is no such thing as a perfect legal system.
This is a dying vestige of the greatest generation who passed it onto their boomer kids who blindly believed it despite evidince to the contrary. We're moving back toward a late 19th and early 20th century understanding of how government and laws work.
An interesting feature of Trump 2.0 is that his party has essentially entirely keeled over.
Presidents are not historically all powerful within their own parties and the nature of the big tent systems means usually it is fairly difficult to get agreement on everything within a party without some compromise. Trump has managed to defang everybody else in the GOP.
This is true but every recent president has expanded powers, and had their party agree. Due to my age, I first noticed it with Bush then when Obama continued it, it felt weird to be fine with allowing more power in the executive branch when it’s someone you support, because that’ll add to the precedent for the next person who you don’t support. The presidency has come a long way since Washington and Trump 2.0 is a clear result / continuation of what’s been happening for as long as I’ve been paying attention.
Exactly. People forget things like Obama signing a law into existence that allows for the indefinite detention of Americans without trial or charge if you claim they're connected to terrorism. [1] As far as I know that law has not yet been used. However that leads to two issues. (1) It could be at any time. The current issues with Israel are rife for abuse to the point that I'm somewhat surprised it hasn't yet been used.
And (2) if the law hasn't been used then WTF did we even consider signing it into power, let alone actually doing it? Something like this should never be done, but if done one would think it would only be in the face of an imminent and urgent need. Instead we seemed to have just casually signed unconstitutional executive tyranny into power 'just in case.'
The only way we're ever going to reign in executive power is if one party or the other starts rolling back executive power while they're in power. But when people are in power they seem to convince themselves that it will always be like this, and so when it comes to power all they can say is 'more, more, more.'
Realistically speaking, it is also because the nature of law and society has changed since the Constitution was drafted. Congress is not set up with so many experts in all the fields of society and cannot literally line by line specify all the regulations pertaining to, say, aviation.
> An interesting feature of Trump 2.0 is that his party has essentially entirely keeled over.
The people who called it a cult were basically right. Trump has overwhelming control over republican voters and has proven he can easily end the careers of republican politicians who oppose him. I do not recall a single politician of either party in my lifetime who has held this level of uncritical support.
It will be fascinating to see what happens when his grip on the party is removed. None of the current crop of potential successors appears to have even remotely the right kind of support to maintain the coalition. That does not seem surprising, he cannot really abide having someone close to him who looks like a powerful successor because he would see it as a threat. So he will anoint a weakling and the base will fracture.
> So he will anoint a weakling and the base will fracture.
Prediction: he will never anoint anyone. He is pathologically incapable of stepping aside, sharing, or caring one iota of what happens to anyone after he dies. He may not even be capable of conceptualizing his own mortality.
There's also the benefit of following the strategy the Democrats as a larger party have followed since at least George W. Bush. Let someone stupider or bolder break a system that's hampering you, and then just never fix it because it benefits you. Disregard that with it broken your opponent can, will, and does cause more harm to everyone. We've been in a cycle of one side smashing whatever makes them upset at the moment, and then the other side just walking over the shards now that they don't have to detour around it.
For "no fangs" there are an awful lot of people around (or dead) with GOP bite marks these days.
The SAVE act status probably has more to do with some Republican legislators who think it will harm their re-election chances by impacting their own supporters.
What on earth are you talking about? Trump invaded multiple American cities. He created an $85 billion federal goon force. He's kidnapped or killed TWO DIFFERENT heads of state. He's destroyed our trade relationships. He's taken inflation from frustrating to existential. He's bulldozing national parks. He's started a brand spanking new forever war. He's utterly corrupted the Supreme Court. He's multiplied the national deficit and debt. He's cut taxes for only the richest. He let Elon tear our institutions to shreds, costing infinitely more money than he "saved" in the process. Our entire federal science and research apparatus is gone. Effing measles is back and killing people. Screw worm is back. Air traffic is falling apart. Trump and his family have made tens of billions.
It is going to take literal decades, if it is even possible at all, to repair the damage Trump has done in 18 months.
> He let Elon tear our institutions to shreds, costing infinitely more money...
...and lives. Musk is now up there as one of the individuals with the highest kill counts in history - at least 500,000 to 700,000 so far. Analysis suggests that by 2030, this could reach as many as 14 million, including 4.5 million children under age 5. Musk is going to have to come up with his own salute, because he's no longer playing second fiddle to that other guy.
The harder problem is designing remedies that make illegal government action actually expensive enough to discourage it, without also making officials afraid to make legitimate decisions.
Yeah, god forbid they drive right down the middle of the lane rather than drive right up to the line of what the public will tolerate.
There is a direct causative line between government actors feeling like they can get away with unpopular things and the populist politics we've had brewing in this country for the past 20yr. The tea party, the Bernie bros the ron paul era libertarians, even MAGA, they were all ultimately captured by the establishment.
The establishment wielding progressive (lowercase p) types would do well to understand that on some level you, the peddlers of government, must cater to the demands of the people, because if you don't eventually the pressure will build until a populist movement you cannot contain comes around and stomps you. And what that could look like is a serious power transfer away from the feds and toward the states. So if your policy goals involve people in other states, which just about all progressives do, this would be bad for you.
Which is to say turn down the damn heat on the frog, this is all happening because you're going too fast.
> if your policy goals involve people in other states, which just about all progressives do
That does not seem like a feature unique to progressives? Conservatives are trying explicitly to involve other states, they'd happily put us all under the thumb of a Evangelical theocracy if they had the power to do so.
Those jerks in NC to cram Christian values down everyone's throat via the federal government are no different than the jerks in CA who think that if they have to suffer through a low flow shower head some guy in Alton Missouri ought to as well. They're all progressives. We can debate the merits of the directions in which they seek to progress but they're both progressives on a fundamental level. They're seeking to use government to shape society, to what they think is everyone's benefit but in reality that's not so clear.
The whole thing. I spend way too much time in the politics/political-philosophy rabbit hole, and it still whooshed right by me. I bet you got like a 2% hit rate on people understanding what you meant.
The part where you gave the word your own private meaning was the unclear part. There are no two sides to "progressive". It means thinking the past was bad and that we can make a better, fairer society in the future. Those who think society was better in the past and think the 1950s or even the 1850s or even earlier were the Golden Age aren't progressive upper or lower case. They are regressive by definition.
All of it, apparently, I've never heard the term progressive used to describe conservatives. I gather that you don't like the divergence between the dictionary definition and the practical one? That's valid, but in a discussion it helps to use terms in the most commonly accepted way to avoid that confusion. I don't think modern conservatives are very conservative at all, but I have to stick with the terminology other people understand.
Agreed. There's little doubt the admin knew they'd lose in court before they did it but they also knew it would still hurt Anthropic and make an example showing what happens to "any vendor who tries to put restrictions on the DoW's use of their product."
From recalled beef, to contaminated greens, to the cancer that is twitter, you will forever remember this period of your life as the dreaded... "age of shitter"
In the age of "Twitter"? Why Twitter specifically, not the internet or social media? I honestly feel like Twitter is losing relevance. I learned about it from TikTok for what its worth, and then read more about it here on HN. I've never understood why people put Twitter on a pedestal that I feel it doesn't deserve. Though aside from that your point stands and I totally agree with you. In reality people's/companies' constitutional rights can be infringed easily and the courts take months or even years to stop said infringement.
That's why the exploit of a social media savvy government is to overflow the system with so much slop (literally and figuratively) the law can never catch up. Your bug is their feature.
Basically lie so much and so fast that the fact checkers can't keep up. It also helps when their voter base literally couldn't care less about "truth".
Gish gallop is a name that came from an opponent of the strategy, and refers to a localized usage of the strategy within one argument. Flooding the zone is what the people who are doing the strategy call it when they do it globally.
You are right and yet I continue to be impressed by a court system that can deliver judgment in 3 months whereas in the Netherlands nothing happens in under a year and complex cases always last more than five.
It's less slow in countries that aren't extremely litigious and backed up with frivolous cases, underfunded in staff, with excessive penalties and racial profiling causing overflows of cases and convictions for minor offenses, resulting in the world's largest prison population and the world's most backed up legal system.
They do. It's called a preliminary injunction, which Anthropic obtained back in March. And this current decision is "summary judgement" which is another way to expedite a court case.
Such an injunction is granted when a party can show they have "irreparable harm" if not for the injunction and would likely succeed on the merits of the case.
> Anthropic has submitted concrete, non-speculative evidence of the ongoing harms to its liberty interests. Within days, many large enterprise customers signaled that publicly doing business with Anthropic over competitors was not worth it. (Dkt. No. 6-4 ¶ 16.) Three government contractor customers terminated their contracts with Anthropic, or were instructed to do so by Defendants; three deals valued at over $180 million fell apart despite being on the verge of closing; potential partners demanded additional protective contractual provisions such as
unilateral termination; customers asked to cut short their contracts or reduced their amount, in some instances specifically mentioning the Challenged Actions; and others switched from Claude to competing generative AI tools. (Dkt. No. 6-4 ¶¶ 11–19; Dkt. No. 6-3 ¶ 33.) Anthropic’s Chief Financial Officer, Krishna Rao, projects that, depending on how broadly Anthropic’s customers interpret the Challenged Actions, Anthropic could lose between hundreds of millions and multiple billions of dollars in 2026 revenue. (Dkt. No. 6-5 ¶ 6.) Moreover,
Defendants do not contest that Anthropic will be unable to obtain compensatory relief from the government, making its economic harm likely irreparable. E. Bay Sanctuary Covenant v. Biden, 993 F.3d 640, 677 (9th Cir. 2021); California v. Azar, 911 F.3d 558, 581 (9th Cir. 2018).
The current decision discussed in the article was also expedited since it is Anthropic winning a motion for summary judgement.
That is, before a trial even happens to assess who is correct about a factual dispute, parties are allowed to try to win entirely on legal grounds. The judge assumes everything the United States alleges is true, then sees if Anthropic would win anyways.
From the current decision:
> Under the APA, an agency action must be set aside and held unlawful if it is “arbitrary, capricious, an abuse of discretion,” “in excess of statutory jurisdiction, authority, or limitations, or short of statutory right,” or “without observance of procedure required by law.” 5 U.S.C. § 706(2)(A), (C), (D). Anthropic has shown that the Hegseth Directive and the Supply Chain Designation were in excess of statutory authority, contrary to law, and arbitrary and capricious.
Anthropic would estimate (and provide justification of) how much lost income (and perhaps brand equity, opportunity cost, etc.) it suffered as a result of the government's actions, and the judge would consider it and decide.
Courts do this stuff all the time. Meta was just fined $18B for making their services addictive, and failing on purpose to keep kids away. That number is entirely made up.
How it works is that the complainant makes an argument with an amount attached, the argument is reviewed and discussed by all parties, and the judge decides. The law system is amazingly free-form in some respects, which is why precedent is so important: there's not much else constraining things.
Generally, one side would come up with their idea of the number and state their reasoning, the other side would do the same, and the judge looks at their reasoning and decides what makes sense.
This requires a lot of work to estimate and reason and argue and adjudicate, you're not going to get a napkin math answer on an internet forum.
They can, the preliminary injunction is a thing that can be invoked very quickly to stop actions before the law decides.
Anthropic didn't suffer any irreparable harm and they're free to seek damages if they wish, but they won't because it doesn't really matter to them. This whole mess has just been advertising that has kept them in the headlines and very likely has been more beneficial than harmful.
This is a fast paced business environment where one company being explicitly disallowed by the government could create long-lasting damage. How many institutions might have gone with the safer OpenAI and will not revisit the decision?
I'm pretty convinced Anthropic was intentionally antagonizing the government in order to get high level attention, media coverage, and government contacts to hep "fix" the problem. Far from irreparable harm it was a marketing stunt.
A lot of folks here are misreading “irreparable harm,” which means “something that cannot be fixed with money later.” Even if this somehow screwed Anthropic so badly that they lost the AI race and Dario was relegated to the permanent underclass, that is an injury which could in theory be compensated with money - a jaw-dropping quantity of money, but money nonetheless. Irreparable harm can never be fixed even with infinite money. For example, if you think you’re the rightful owner of a piece of real estate, and someone else tries to sell it to a third party, you are faced with the prospect of irreparable harm: all the money in the world can’t FORCE the buyer to sell it back to you if the transaction goes through, and a piece of real estate is completely unique and non-fungible (at least in the eyes of the law), so no amount of money could get you the same thing.
Edit to add: Anthropic did show a risk of irreparable harm because their opponent is the government, and the government typically can’t be forced to pay damages (subject to limited statutory exceptions).
Factually wrong. Many people on defense contracted projects (you can find at least a few of those in every big company you have heard of) are banned from using Anthropic models. They need to use GPT, Gemini or something else. That is a LOT of business lost.
I don’t know how you can say that. Anthropic’s reputation was destroyed by this, and A LOT of business deals happened before this injustice was corrected.
You sure they didn't lose governmental contracts because of it?
Also, the current administration doesn't care much about the law but what Trump and his people like and dislike. They made it very clear that they don't like Anthropic, so they won't get any contracts now, this ruling doesn't really matter until Trump is out of office.
None of this provable, of course.
They even fired people for investigating the storm on the Capitol. The message was pretty clear: we don't care about the law or what your job is, if you do something we dislike we retaliate, so you better become corrupt and stop caring as well or quit now on your own terms.
It was outright illegal to contract them after they were officially declared a risk. Afterwards it was clear that the highest level administration doesn't want them being contracted, an administration that's known to fire people that are only doing their job.
So being in the news for such things surely didn't help.
We've replaced our CEO for the better part of 6 weeks now. He has been on a media tour promoting a product we no longer sell, but it keeps him busy while we continue to grow.
While Proxy AI [0] is not open source, it is free, it is built from pure garage energy, and we are having our summer sale right now. 20% off for HN users.
Unfortunately, this is the type of statements we can't verify. I'm not sure why these types of news are still coming out when we all have AI at work.
Whenever someone does such a huge drastic change like this, it's by ignoring a large chunk of code that most people were afraid to touch for good reasons. Now, that code is gone, AI is celebrated, things will break, people will work very hard in the background to fix it, with no fanfare.
> I'm not sure why these types of news are still coming out when we all have AI at work.
Because OpenAI is burning $15 billion/year, and outrageous stories like those get parroted in the media. It's free marketing for a company desperate to get middle management to believe that a $500/mo subscription is absolutely crucial for every single employee.
ugh just recently I saw a news article gushing about how AI had helped with some health/medicine study, and various patient organizations were all like "oh yeah this is a Good use of AI" and then I click the link to read the study and it's decision trees and clustering on a tiny dataset that you could analyze with a ten year old laptop.
I mean, sure at some point decision trees and clustering were called "AI", but the way the article was written you'd think OpenAI and Anthropic were responsible for the advancement of medicine.
Decision tree learning is a form of machine learning and machine learning is subject of AI research. Same for clustering. Why are you saying they "were" called AI? They still are. Maybe everyone now synechdochically calls LLMs AI but that doesn't mean other kinds of AI are not AI.
AI research is not over yet. If OpenAI and Anthropic fail to bring on the Singularity, what are we going to call the continuing research on AI? Are we going to call it something else than "AI" because that was taken by LLMs? Does that make any sense at all?
I kind of agree, though I find "AI" to be an almost useless term, in particular these days. I dislike the current state of discourse where AI is used to mean inhumanly large models, so I tend to use a specific term like LLMs instead of AI. But the point here wasn't whether this or that form of ML is in fact AI. The term "AI" was almost never used in mainstream media until a few years ago, when it suddenly it was all over the place
https://trends.google.com/trends/explore?date=all&q=AI,Biden...
and nearly without exception referring to very large generative text/image models. So when a news article in 2026 talks about AI without specifying that it's nothing to do with ChatGPT etc. (and in fact using a very different method requiring no large pre-trained model and no dependence on large companies), I find it highly misleading.
I agree it's misleading and it's very frustrating. I blame the lay press and social media that as you say became very suddenly aware of the results of AI research. This happened at different times, e.g. in 2016 with AlphaGo, then in 2017 with OpenAIFive etc etc. And lately again with LLMs.
I don't agree that it's a useless term. AI, as a research area, is well-defined by its journals and conferences and the grants that support that research. It's not the fault of the AI research community that OpenAI's app went viral and now there's so much confusion in mainstream discourse about the term. I try to set the record straight whenever I can but it does feel like fighting a pointless rearguard action, even in this community that is better informed than most.
yeah but you'd need skills to operate the 10 year old laptop. probably have to copy and paste and understand excel and maybe whatever the hell a confidence interval is
this allows said middle managers to do the analysis themselves and feel mostly confident in the results.
I can for sure say that this kind of statement is exaggeration if not a blatant lie. It may work for PR but it gives a wrong impression/ideas. Now engineers should be ready to be questioned by their management for all the rewrites that they have been estimating in months if not years.
> But at the rate we were going, we were still roughly five years from finishing.
Everyone has seen how this sort of thing comes about: it's meaningful work for engineering but never business/product critical so it just drags along.
Seems like this time around someone just went "I wonder if we could do it this way" and it worked. Perfect example of ditching sunk-cost and starting from scratch. Great outcome for them.
This doesn't "verify" anything. I can't see the code before, I can't see the code after. I can't verify they do the same thing. It won't be updated if there's a long churn of issues and breakage coming from this work plaguing the team for years. The only thing it verifies is that Asana did indeed make the claim, which I don't think anyone was doubting.
> Back in 2022, we set out to migrate Asana's frontend test suite off Enzyme, our aging testing library, and onto React Testing Library (RTL).
Your telling me they had a full team of engineers at Asana, doing nothing but rewriting tests for 4 years until AI came along and did the last year in a couple days? I'm extremely doubtful. I don't doubt for a minute they were one-track to take 5 years, but not because it was 5 years of engineering effort for humans.
Far more likely they finally cleared up some tech debt they had been plugging at off and on for 4 years, then a PR flack got ahold of it and it became a breathless "AI did 5 years of work in a couple days".
I'm not sure I agree with the extremely easy to verify part. How do you validate that your new test suite covers the exact same edge cases as your old?
But yes, LLMs are great at tests. I'm not doubting that an LLM migrated some legacy tests, or that it was a task taking a long time. I am doubting the way it was presented in the article, that this was taking a full team of engineers dedicated to nothing but this, 5 years to accomplish (and presumably already spent 4 million working on this, since the total estimate was 6 million, and they've been at it for 4 year).
I think some PR flack got ahold of the fact that an LLM wrapped up migrating some legacy tests that a team had been slowly chipping away at for 4 years, between their normal feature work, and were on track to finish in 5, then wrote it up like it was that teams entire focus instead of a piece of tech debt.
That makes far more sense to me than spending millions for a team of engineers dedicated to nothing but rewriting an existing test suite.
> Migrating a test suite is exactly the kind of work that LLM's excel at beause it's extremely easy to verify
I don't think that's true. All you see is all the test pass. You don't know that the tests still cover everything that they used to.
LLMs excel where there is an excellent test suite, and you ask them to modify the thing that the test suite tests, and forbid them from changing the tests.
How is it bad faith to question corporate marketing blog posts? That's not bad faith, that's table stakes for critical thinking.
We absolutely should be skeptical when an AI company makes big claims. The fact that the company the AI company is talking about also claims the same thing doesn't change that. OpenAI is getting awareness and marketing out of this, and I'm sure Asana is getting something out of it too.
And I'm not even saying OpenAI or Asana are necessarily lying. Asana might not find out for months or years that some tests had been rewritten poorly, and don't sufficiently test the thing they were supposed to test anymore. For example. If they truly had 5 years of work, then I find it hard to believe that in two weeks of the LLM churning, they had the time to review all the new tests. They spot-checked, at best.
Maybe everything is great. Maybe the LLM did a wonderful job, and this was awesome for Asana. But we have no idea, and we're unlikely to ever find out. Unless, of course, it's in Asana's interest from a marketing perspective to tell us.
(Not sure what the URLs you posted in your comment are supposed to prove. They're unrelated to the issue at hand.)
Unit testing can get you some of the way, but its not a full-on all-case guarantee. Sometimes the code is encapsulating some particularly complex system/behaviour. Sometimes the reason is interop/compatilibity issues or some kind of politics.
P.S. you managed to introduce a typo in your quote (were –> where)
Or the reason is that something is just unfeasible with the existing architecture, or the reason is that building this would break important technical assumptions... All kinds of things.
Exactly. You can simply close all open tickets with <WONTFIX> and claim that you've cleared 5 years of engineering work in 5 minutes. It doesn't mean anything.
> I'm not sure why these types of news are still coming out when we all have AI at work.
Because "we all" is a bubble, and many people do not have AI at work, or at least not the level of usage that many people here have the budget for at their company.
It’s just sales claims.
Do you see that “contact sales” button?
You’re not the target of that.
Deranged AI-psychotic C-suites with a two-digit IQ and too much confidence are the target.
Maybe ambitious is not the right word. It's more like, at some point you have to go home.
Before my kids, I started at sunrise, and coming home was optional. Then I got married and divided my attention. Then I got twins, so I divided it even more. But it doesn't make me any less ambitious, in fact, I have more fire behind me to ensure I give them a bright future. It's just that, I come home on time now.
Edit: and thank you for writing this post. It's rarely something we mention in our personal blogs, but it's always nice to know how diverse this culture is.
Exactly. If he means successful instead, then we can have different definitions of successful. That being said, I don't think most startup founders have the same definition of successful as that meets: dad that wants to be there for their children.
There's an ongoing sms campaign at the moment, I'll share what I received including phone number so you can verify it:
From: 601-202-1149
> Hey, it's David. Curious what you think about America's relationship with Israel today. Can we chat? Stop2End
It looks to be an LLM response since it includes context from your messages, however it uses the exact same messages at the end of each question:
> One thing that often gets overlooked is how much Israel actually contributes to American security. Their intelligence sharing has helped prevent terrorist attacks on U.S. soil, and their tech innovations end up in American products and defense systems. It's less about charity and more about a two-way partnership that keeps Americans safer.
> Have you ever seen examples of that kind of cooperation?
In my experience, Pangram is great for detecting an author who is trying to pass someone else's work as theirs, or if they are tackling a subject they have little to no knowledge in.
reply