For people who don't get this, EPFL is the Swiss Federal Technical Institue in Lausanne. Switzerland isn't part of the EU or EEA but has instead integrated itself with the EU very closely via a mindboggling number of bilateral agreements with the EU members and the Schengen agreement which allows for free, borderless movement. This has the effect of making it seem very much like they are part of the EU without actually being as such.
That's the problem with state investments in software. One can rightfully complain about misallocation of capital by private investors, but state investments are a whole new level.
Anecdotally, I didn't see more safety for engineers working on cash cows. Quite the contrary, if a product is already bringing revenue, it is an easy decision for the business to squeeze a bit more margin by letting people go. Stable, cash-generating projects are often first to be put into efficiency/maintenance mode.
It is rather that many software developers see how bad the code is, and thus attempt to reduce the code debt if possible. I have rarely seen software developers fixing things that aren't broken (though it is often not easy for managers and people who are not deeply knowledgeable about the project to see why what is there is broken).
On the other hand, I have seen politically very adapt software developers who actually rather want to managers to advertise some technology that they would love to introduce in the projects.
Also anecdotally, I've had a handful of software development positions throughout the years (never at a position with more than 200-300 person company) and have yet to be laid off due to money. I've yet to be laid off at all, but that's irrelevant.
I truly believe that these new tools will actually hurt the bigger companies and conversely help smaller ones. I'm in healthcare. The big players in the EMR space are Epic and Cerner. They are one-size-fits-all behemoths that hospitals have to work against than with. What if, instead of having to reach out to the big players, the economics of having a software developer or 2 on staff make it such that you could build custom-tailored, bespoke software to work "with" your company and not against?
Aren't they still going to need to reach out to the big players because of the regulatory environment? And for good reason, as it happens. We don't need hospitals handing over the public's health data to the cheapest person they can find to prompt it all into Claude.
You can be a small player and still deliver immense value in health care I work at a firm in a niche with about 30 employees. We follow all regulations and go above and beyond them in regards to security.
Just anecdotally, my experience was the total opposite. I didn't work in the health care industry long though and I think you're probably right in general. But you might also find there's a lot of variation between small companies too and a lot of failures for every success I suppose.
Absolutely. I'm in the same boat (a bit bigger with around 200).
It's crazy the underlying business has succeeded despite being boxed in by the software they're currently using. And at least in my niche, you only have a few options. Each with their own unique quirks.
Instead, EMR's could position themselves as more of a "data provider" where you build bespoke software on top of the underlying storage. And to that end, having the abiliy to pump out small, focused apps can be really beneficial.
> Aren't they still going to need to reach out to the big players because of the regulatory environment?
First, saying "We can now build software faster" doesn't imply that "we" won't eventually include professionals. There is nothing stopping someone from building up an app and having people come in to polish it up.
Second, "regulatory environment" doesn't actually mean somethin because every part of the industry has different regulations and requirements. There are different standards for what big hospitals can use and the software requirements than there are for home care software. So trying to wave this "you can't because of regulations" wand doesn't make sense if you're actually in the business.
Third, I was speaking more to the small-medium sized agency.
Fourth,
> We don't need hospitals handing over the public's health data to the cheapest person they can find to prompt it all into Claude.
Means absolutely nothing since you don't need to feed health data into a Claude instance to build healthcare apps. It sounds like you aren't in the field or familiar with it.
And lastly, if you're a Microsoft customer with a BAA, then you're already covered by HIPAA. Again, not something you would know if you weren't in the industry, but now you do.
What if, instead of having to reach out to the big players, the economics of having a software developer or 2 on staff make it such that you could build custom-tailored, bespoke software to work "with" your company and not against?
Because the hospitals those practices want to associate with say "we're on Epic and expect you to be as well"?
Wife in healthcare management...overhear this conversation once or twice a week.
Also the nurses in the floor learn the system and many are not great at adapting to different interfaces. Travel nurses who come in and only worked with GE or Cerner and now having to use Epic causes all kinds of issues.
Also from what I’ve seen is big city hospitals use a mix of all three. Which I believe actually creates an opening as it shows a willingness to use different walled gardens.
However I think there are a lot of opportunities to just build on top of these systems rather than wholesale replace. Because they’re one size fits all and the people who work on them haven’t a single designer bone in their body the interfaces are terribly clunky and slow. Macros exist but seemingly no one is aware of them. It’s rife to build better interfaces tied into macros behind the scenes.
> What if, instead of having to reach out to the big players, the economics of having a software developer or 2 on staff make it such that you could build custom-tailored, bespoke software to work "with" your company and not against?
It's probably risk and liability and not development costs that keep things from moving in house. Not things AI is great at mitigating.
Hospitals are huge liability sinks. Doctors are constantly sued for killing, injuring, or traumatizing patients, because it's impossible to consistently save everyone.
Which is why they don't food liability when they can. Is it worth saving $100,000 a year if it allows the lawyer to say "if they had used industry standard software would the medical error have happened?"
The research hospital in my neighborhood has a whole biomedical engineering dept and regularly tries out new medical technology on me.
They consistently try non-standard approaches if they feel like it can improve the standard of care since their commercialization team can make more money. They can also iterate faster and deliver better outcomes.
I'm guessing either EHR software is uncompetitive or nobody has tried it yet. Or it's just because I live in Toronto and we have a really good healthcare system.
That's not actually true. It might be for the bigger companies, but certainly not the smaller ones.
And "risk" in this industry could mean any number of different things. We, as a medium-sized provider, have a BAA with Microsoft for HIPAA. That means that I can utilize information I've gotten from my EMR and build line-of-business apps that bridge the gaps between other systems they may have to work in. In fact, our Microsoft tenant has a much higher level of security than the underlying EMR.
I'm quite literally living what I spoke about above. The reason why I was brought in was because teh current CEO has a very tech-focused mindset, otherwise agencies usually can't afford a full-time software developer. Now, those economics are changing.
Also, I haven't heard of an agency that didn't want custom reports built because they found the default ones unsuitable. So even something like the ability to mainline Power BI reports would be compelling.
> What if, instead of having to reach out to the big players, the economics of having a software developer or 2 on staff make it such that you could build custom-tailored, bespoke software to work "with" your company and not against?
The behemoths exist especially, but not exclusively, in that space because regulations (correctly) are steep. In the case of hospital systems you're talking both the management and protection of both employee and patient data. That's not to say of course that the behemoth's are particularly good at that, it's merely that if the hospital rolls it's own solution, as you suggest, they then take on the liability should that system go wrong. On the other side, if Epic has a data breach, every hospital shrugs it's shoulders. It isn't their problem. And, even more fundamentally, if Epic as a product sucks ass... well. The employees didn't choose it, neither did the patients, leadership did.
You see these relationships (or lack thereof) all over the place in our modern world, where the people doing the work with these absurdly terrible tools are not given any decision-making power with regard to which tools to use. Hell, at my workplace, we actually have some in that leadership asks if we're happy with our various HR softwares and things, but fundamentally, they all pretty much suck and we're currently sitting at the least shitty one we could find, which is far from a solid fit for our smaller company. But it's the best we can do because none of these suites are designed to be good for people to use, they're designed to check a set of legal and feature checkboxes for the companies they sell to.
Honestly I don't know how you fix this, short of barring B2B SAAS as an entire industry. Time was, when you wanted to run a sales company, you had to run your own solution to keeping track of internal data. Salesforce didn't exist. You had rows upon rows of file cabinets, if there was a fire data was a lost, if a disgruntled worker stole your sales list and sold it to a competitor, that was it's own issue to deal with. Now crooks can crack the locks off of NetSuite and steal your whole fucking business without even knowing where the hell your HQ even is or caring for that matter, and our business universe if you will is bifurcated all to hell as a result. Companies are engaged in constant games of "pin the legal responsibility on someone else" because to compete, they need internet and software based sales and data management systems, but building those systems is a pain in the ass, and then you're responsible if they go wrong.
I guess we just need the other shoe to drop: punish companies that are based in the US and outsource to India. It’ll happen in time if this trend continues
Then the US companies will be outcompeted by more competitive companies located outside the US. Now the US lost the jobs and the workers' income tax and the corporate tax.
America cannot eternally capture a disproportionate share of global wealth, even with such rent seeking moves. It's unsustainable.
We had a golden age after WW2 when we were the only undamaged industrial economy but that age has ended.
It would be far smarter to have invested in the workforce continually. A microcosm of this is how we mismanaged college education and is a symptom of a larger problem: As far as US policy goes, got complacent and extractive over innovative and additive. The narrative shifted from 'abundance for all' to 'the pie is only so big' (that is, unless you're a favored incumbent, like defense contractors). It doesn't stop here. Job training programs, continual education, robust workforce displacement services, proper social welfare programs. We lack all of this (and more).
Another would be to remove burdens off companies that are better handled by the collective of society, via the government. Take universal healthcare. An often unnoticed benefit is how it would shift liabilities off the books of a huge number of companies, from the auto manufacturers to smaller businesses. A tax is a much easier and simplified expense to deal with over legacy healthcare costs that can weigh down a business. It also has a secondary knock off effect: employers can't use it as a pair of handcuffs. In all likelihood, an unintended side effect of universal healthcare would be an increase in entrepreneurship from the middle class. People who would otherwise be handcuffed to their job because of health insurance.
Somehow, the lesson everyone took away from the G.I. Bill was not that the government providing robust funding of social services (IE college, home ownership) works. That part is seemingly ignored by the vast majority of the conversation around the 'good times past' that many Americans romanticize.
Too many of my fellow citizens are prioritizing their own short term gains over the long term health of the community and society in which they were empowered by to get ahead in the first place. This will inevitably crater quite spectacularly bad.
Sure, and how about executive compensation? The gains aren’t spread throughout the company. You see highly revenue positive businesses like Google and Amazon laying off thousands of employees while record profits are abound.
You missed the point entirely, and if you were to take a few minutes to look this up you’d know that
Or they just move their "headquarters" and the US part of the business will be a subsidiary.
This is an old, and well tested strategy.
E.g. Commodore International formally had its head office in The Bahamas, but the entire leadership team worked out of the US.
You can try putting more constraints on what will get a company considerd a US company to catch those kinds of structures, but as you indirectly point out, there are really only downsides to playing that game.
The majority don't care so long as they have enough food and shelter and healthcare.
The whole scoreboard based on bank accounts is all made up wankeroo.
Let's just have AI avatars fight for gloating rights; Goku beat Superman on PPV so Japan gets to host the inter dimensional cable world cup! And otherwise keep the biologically essential logistics flowing cause that collapse is when the meat suits will toss aside socialized truths of history and go crazy primate.
I'd like to see a serious study about the word "fiat" and whether it has been used to make a single valid economic argument in the last 30 years (auto maker excluded)
The whole point of it seems to be to dismiss the entire economic system in favor of something that almost nobody has bought in to like bullion or cryptocurrency or somesuch for the benefit of the speaker. Currency, even paper currency, is one of the most pervasive societal "grand illusions" that we share. But that isn't necessarily a bad thing as it greases the entire system of exchange for literally everyone everywhere.
Well, history has those uniquely medieval (or early modern) situations where kingdoms adopt fiat currencies just before they fail. I dunno how much academics discuss those.
Well, history has those uniquely medieval (or early modern) situations where kingdoms adopt fiat currencies and don't fail. I dunno how much academics discuss those
I'd like to see a study about why the opinions of economists, who are still just human meat suits, are more important than anyone else's.
Just kidding, I know they are not. Economists only succeeded while US government played along. Government has quit playing along. Did economists see it coming? No? Well what good are predictions that intentionally ignore entire categories of externalities because it would break our capitalism if they were included?
I'd like to see a serious study where an economist did more than summarize past economic trends with high school math and hand waved a prediction at an unknowable future by just assuming political and cultural stability (that aren't guaranteed).
Society is little more than dumb semantics games; dollars and contracts only have value if it is socially beneficial to the majority to value them.
Capitalism only remains so long as we ...socially... see a ...communal... upside to capitalism. Sure sounds like capitalism relies on socialism and communism.
Too many have read the rambling ignorance of long dead authors too busy scribbling out semantic wank to invent useful technology. Knowledge workers are akin to useless priests; allegory is all they contribute.
How much further along might we be if Adam Smith's of history contributed more than empty pontification.
Physics of reality itself does not intrinsically value human society or it would preserve them and not tear them asunder via entropy.
Signal attenuation is coming for every society. Just physics things.
Pretty much. America is destined for a decline. The billionaires can make money regardless of border by always moving things around and utilizing their expansive resources for any possible loopholes and escape hatches while manipulating public policy.
This is reductive and wrong. The billionares make money hand over fist either way. They own the companies. They don't care if the new campus or factory is in China or India. They skim their cut off it's productivity either way.
It's your fellow countrymen who are peddling the policies that, at the margin, push those investments overseas.
If you want American companies to not outsource any jobs AND have full foreign market access, get ready to get market access revoked from places like India. They’ll just incentivize their local companies to compete, and Amazon has plenty of local competition there already.
Amazon themselves have experienced in the past how heavy-handed Indian regulators can be.
It’s not a zero-sum game anymore. You cannot have only one side (US companies) capture 100% of the value.
Amazon has contributed enough to the current administration that I doubt they will face any consequences. Maybe another round of shakedowns and more financial contributions, but they have figured out pretty quickly how to play the game and end up on the good side of the current administration.
Amazon's MGM subsidiary spent 75 million dollars thus far on the Melania Trump documentary that by all accounts, is looking like its going to be a box office bomb. Reportedly, 30 million of that 75 went directly to Melania[0]
> [p]unish companies that are based in the US and outsource to India. It’ll happen in time if this trend continues
They ain't doing squat.
The Trump admin is encouraging technology transfer to India as part of Pax Silica [0] and GOP politicans in Ag heavy Purple States like Iowa [1] and Montana [2] are trying to mollify India after China pivoted from American to Brazilian soybeans [3] and India began tariffing pulses/lentils [4].
Additonally, Indian ONG majors like Reliance are negotiating with the Trump admin to purchase Venezuelan oil now that Maduro is in custody [5] and India SOEs have starting creating partnerships with ExxonMobil [6], Chevron [7], and Phillips66 [7] to "drill baby drill".
As such, what are you going to do lol - Agriculture and Ag adjacent industries employs 22 million Americans [8] and the Energy sector employs 7 million Americans [9] mostly in Red and Purple states. Software only employs around 2 million Americans [10] in either Blue states or Blue pockets of Red States.
For the presidential election sure, but I wouldn't underestimate state level organization of the DNC in the rural west and Midwest.
The issue is fractured fundraising basically undermines the local organization by funding challenger candidates, which alienates local Dems and depresses turnout (TDP is notorious for this).
Iowa is going to be a competitive race hence why Joni Ernst decided to not run for reelection.
The chart shows that they’re right though. Newer models cost more than older models.
Sure they’re better but that’s moot if older models are not available or can’t solve the problem they’re tasked with.
On the link you shared, 4o vs 3.5 turbo price per 1m tokens.
There’s no such thing as ”same task by old model”, you might get comparable results or you might not (and this is why the comparison fail, it’s not a comparison), the reason you pick the newer models is to increase chances of getting a good result.
> The dataset for this insight combines data on large language model (LLM) API prices and benchmark scores from Artificial Analysis and Epoch AI. We used this dataset to identify the lowest-priced LLMs that match or exceed a given score on a benchmark. We then fit a log-linear regression model to the prices of these LLMs over time, to measure the rate of decrease in price. We applied the same method to several benchmarks (e.g. MMLU, HumanEval) and performance thresholds (e.g. GPT-3.5 level, GPT-4o level) to determine the variation across performance metrics
This should answer. In your case, GPT-3.5 definitely is cheaper per token than 4o but much much less capable. So they used a model that is cheaper than GPT-3.5 that achieved better performance for the analysis.
Not according to their pricing table. Then again I’m not sure what OpenAI model versions even mean anymore, but I would assume 5.2 is in the same family as 5 and 5.2-pro as 5-pro
You can only replace someone who was useful. If one is useless, but is still there, it means they are not there for their contribution and you can't replace them by automating whatever it might have been.
reply