You wrote: "Isn't it funny how the Republican Party always gets very concerned about spending and the reach of government when the Republican Party doesn't control government; but just as soon as they do have control they start spending like crypto bros and use government to interfere in literally everything that doesn't fit their questionable narratives?"

See also: "The GOP used a Two Santa Clauses tactic to con America for nearly 40 years; This scam has been killing wages and enriching billionaires for decades"
https://www.salon.com/2018/02/...
        "The Republican Party has been running a long con on America since Reagan's inauguration, and somehow our nation's media has missed it - even though it was announced in The Wall Street Journal in the 1970s and the GOP has clung tenaciously to it ever since.
        In fact, Republican strategist Jude Wanniski's 1974 "Two Santa Clauses Theory" has been the main reason why the GOP has succeeded in producing our last two Republican presidents, Bush and Trump (despite losing the popular vote both times). It's also why Reagan's economy seemed to be "good."
        Here's how it works, laid it out in simple summary:
        First, when Republicans control the federal government, and particularly the White House, spend money like a drunken sailor and run up the US debt as far and as fast as possible. This produces three results - it stimulates the economy thus making people think that the GOP can produce a good economy, it raises the debt dramatically, and it makes people think that Republicans are the "tax-cut Santa Claus."
        Second, when a Democrat is in the White House, scream about the national debt as loudly and frantically as possible, freaking out about how "our children will have to pay for it!" and "we have to cut spending to solve the crisis!" This will force the Democrats in power to cut their own social safety net programs, thus shooting their welfare-of-the-American-people Santa Claus. ..."

So it is not hypocrisy so much as a precisely-thought-out effective political strategy. Whether the majority of voters in the USA like the results or realize where those results come from is a different issue.

It has seemed to me, for a very long time, that modern AI systems would need to be integrated with standard RDBMS systems for reliable persistant storage of raw information, some sort of no-sql database (memcache or some variant) for persistant storage of associations, some sort of document database for blocks of textual information, a SPARQL system for searching semantically-marked information within the document database, and a more old-fashioned back-propogation NN to provide a store of understanding that the user can directly manipulate.

Probabalistic classifiers are all fine and good, but only for a subset of the tasks needed. The above structure is a very loose, wildly-speculative initial framework. It's almost certain that if you actually tried building an integrated multi-model system, that you'd end up making a lot of changes to this basic idea, but that you'd end up having to implement the same core concepts that are identified in it.

The British government, in a desperate bid to increase profitability, has trademarked Alan Turing himself.

How is ChatGPT supposed to know the difference? Both involve hands and mouths, so clearly it's the same context. :P

From their careful selection of text, they WANTED it to mean something else so badly that they couldn't handle putting in the full text. It's a common blight on today's Internet, where people want other people's writings to mean something other than what was meant by the writer, so carefully select the words they read.

That's the entire point. Trying to solve other people's problems NEVER WORKS. You CANNOT control others into responsible behaviour, but you CAN place them in a position where they will choose to be responsible of their own accord. It is the ONLY way that works. It is the only way that has ever worked. If you look at computer programming, you will see this repeated over and over - well-meaning "hard rules" are ignored, STANDARDS are kept.

You must give them parameters and force them to find their own solution within those.

I just wanted to add that whatever the truth there, this idea that LLMs are not (by themselves) the way forward is increasingly appearing in various places. One recent example on Slashdot:
https://slashdot.org/story/25/...
"Project Prometheus is building AI systems that learn from physical experiments rather than just analyzing digital text."

Humans learn to speak usefully with just a few years of immersion in a social world and without reading the entire internet. My college advisor back in the 1980s (George A. Miller) though this suggested language had a partially genetically-wired component in the brain even as much was also learned.

Beyond reading Asimov robot stories as a kid, I first learned more formally about AI taking an independent study course in High School in the late 1970s based around Patrick Winston's first edition Artificial Intelligence textbook.
https://en.wikipedia.org/wiki/...

Then in the 1980s, some of my college work was also related to AI as cognitive science and exploring triplestores and so on (which very indirectly helped inspire George to create WordNet as I was graduating, where WordNet lead to Simpli and Google AdSense). I spent about a year hanging around the CMU Robotics Institute after graduation (where I got to ride in the first "Autonomous Land Vehicle" or "ALVAN"). And then I was a research assistant co-managing a robotics and expert system lab for a time. I also made one of the first simulations on a Symbolics of kinematic self-replicating robots (presenting that work at a conference on AI and simulation, where I commented on the total surprise to me when I saw emergent behavior of unexpected cannibalism of offspring in it until I kludged in a virtual sense of smell to avoid eating creatures that smelled the same). As a grad student later I learned a bit about neural networks related to self-driving vehicles.

I later worked for a time in IBM's speech research group in the late 1990s (mainly using existing tools to build implementations, aspects of which were forerunner to Apple's Siri as IBM's "Personal Speech Assistant" and also an interactive speech-operated display wall I built mostly for fun which was intended to in-theory eventually support advanced design and also patent writing).

Anyway, with that for context, I think LLMs are pretty amazing, but they just don't seem like how humans learn to think and speak. Not saying they can't be useful as part of a larger system though. But fundamentally, even if neural networks are involved, humans think in concepts (or word senses, as in WordNet) which they mostly learn by inference from just a relatively few examples. And that learning tends to have a precise aspect to it related to the actual experience and some notion of "truth" (as in actual experience even if the experience is hearing or reading about what someone else experienced or said they experiences).

So the idea proposed here by "Cringely" makes some sense (as part of this trend to seeing the limits of LLMs) -- although whether or not he can pull it off is a different issues.

But there remains a concern of whether or not such a thing (making powerful self-taught AIs) is worth doing right now given a competitive economic system and also the existential risk of creating essentially a new intelligent species (one without all the evolved safeguards humans have as a social species, limited as they may be as demonstrated by various tech-bro behavior). Anyway, such concerns is why I mostly left the AI research field in the 1980s (other than to kibitz about it from the outside).

This YouTube comment was not posted by me but it almost could have been in some ways:
https://www.youtube.com/watch?...
      "@Jenkkimie 2 weeks ago
      Former AI developer here. Hear Mo Gawdat's message to heart. I regret my past, regret that I helped companies to build AI's at all. I can't undo history but I left the AI industry when I saw companies were starting to plan on using AI in unethical ways that I could not stand by. I've lost a lot of money over the years but as far as I am concerned that is the sacrifice I made because I don't want to be part of the destruction of humanity and the world.
      There has got to be better ways to use AI than pure greed, and we need to do better than this. To remember ethics, not just our bank accounts. So I've joined among many other former and current AI developers in advocating for regulations, change of how we think about economies and the role of money in our world and what is our place in it. Maybe we are fighting a losing battle but all of us should do what we can to steer and orient this world to a better tomorrow rather than submit to the will of the oligarchs evil desires. The fight is not over yet, we can still change the direction of it all."

Mo Gawdat (interviewed in the video that comment is posted on) is the only major AI executive who so far I see seems to get the main idea my sig in relation to AI: "The biggest challenge of the 21st century is the irony of technologies of abundance in the hands of those still thinking in terms of scarcity."

Whatever AIs we build, unless we (or they) understand that irony, it seems unlikely that there will be a happy result for humanity of such work.

So, "People With Disabilities Don't Exist" then?

My father was recently paralyzed by Guillain-Barré, so I'll let him know, thanks.

Wow. Thanks for posting this, A.C.. In trying to verify any of what you posted (which was all news to me), I found this:
"The cost of lies: A Mineserver story" by Jeremy Reimer
https://jeremyreimer.com/rocke...
      "Creating and shipping a brand new product is insanely difficult. It takes a ton of money, sweat, and time. Even people with tons of experience can underestimate timelines and encounter unexpected difficulties. So telling the story of a failed Kickstarter is not especially interesting.
      This is not that story.
      This is a story about what happens when someone builds up a reputation over decades of work and then destroys it in a couple of years. Not because they failed, but because they lied about it. Over and over again. Until the lies got too much to handle, and they had to create newer, even larger lies to cover them up.
      Why would anyone do this? We'll get into that at the end. But first, the story. ..."

I can still wonder on the use of the word "lie" in that article by Jeremy Reimer versus, say, "irrational exuberance" especially if his kids were involved in making the Minecraft server project happen? But the article does make it sounds like a larger pattern. Ironically, the behavior even sounds a bit like an overly-people-pleasing LLM hallucination?

Having read many Robert X. Cringely articles in InfoWorld and so on way back when, I would be sad if this was all true. Kind of like losing faith in a celebrity of computing from my younger days.

Related (although in general I have not found it that true about most computing people):
"Never Meet Your Heroes: What It Means & If You Should Meet Them"
https://www.wikihow.com/Never-...
        "Itâ(TM)s a proverb that suggests meeting your idols can lead to disappointment. âoeNever meet your heroesâ is a piece of advice that means people shouldnâ(TM)t meet their heroes because they may be disappointed by the heroâ(TM)s true personality. This happens because people tend to idealize people they look up to instead of viewing them as multifaceted humans with flaws, and they may have unrealistic expectations about what will happen when they meet their hero.
        The hero might not have the time, energy, or interest in meeting their expectations, destroying the perfect image that person has built in their head.
        The logic behind this proverb is that many celebrities craft public personas, and the image they portray online or on camera may be vastly different from how they act in real life.
        With that being said, some people say that meeting your heroes can be a positive experience and serve as a reminder that heroes are no different than normal people. ..."

Related by me from over two decades ago: "An Open Letter to All Grantmakers and Donors On Copyright And Patent Policy In a Post-Scarcity Society"
https://pdfernhout.net/open-le...
        "Foundations, other grantmaking agencies handling public tax-exempt dollars, and charitable donors need to consider the implications for their grantmaking or donation policies if they use a now obsolete charitable model of subsidizing proprietary publishing and proprietary research. In order to improve the effectiveness and collaborativeness of the non-profit sector overall, it is suggested these grantmaking organizations and donors move to requiring grantees to make any resulting copyrighted digital materials freely available on the internet, including free licenses granting the right for others to make and redistribute new derivative works without further permission. It is also suggested patents resulting from charitably subsidized research research also be made freely available for general use. The alternative of allowing charitable dollars to result in proprietary copyrights and proprietary patents is corrupting the non-profit sector as it results in a conflict of interest between a non-profit's primary mission of helping humanity through freely sharing knowledge (made possible at little cost by the internet) and a desire to maximize short term revenues through charging licensing fees for access to patents and copyrights. In essence, with the change of publishing and communication economics made possible by the wide spread use of the internet, tax-exempt non-profits have become, perhaps unwittingly, caught up in a new form of "self-dealing", and it is up to donors and grantmakers (and eventually lawmakers) to prevent this by requiring free licensing of results as a condition of their grants and donations. ...
          Consider this way of looking at the situation. A 501(c)3 non-profit creates a digital work which is potentially of great value to the public and of great value to others who would build on that product. They could put it on the internet at basically zero cost and let everyone have it effectively for free. Or instead, they could restrict access to that work to create an artificial scarcity by requiring people to pay for licenses before accessing the content or making derived works.
      If they do the latter and require money for access, the non-profit can perhaps create revenue to pay the employees of the non-profit. But since the staff probably participate in the decision making about such licensing (granted, under a board who may be all volunteer), isn't that latter choice still in a way really a form of "self-dealing" -- taking public property (the content) and using it for private gain? From that point of view, perhaps restricting access is not even legal?
        Self-dealing might be clearer if the non-profit just got a grant, made the product, and then directly sold the work for a million dollars to Microsoft and put the money directly in the staff's pockets (who are also sometimes board members). Certainly if it was a piece of land being sold such a transaction might put people in jail. But because the content or software sales are small and generally to their mission's audience they are somehow deemed OK. ...."

Relevant: https://en.wikipedia.org/wiki/...
"In the US, SMPTE is a 501(c)(3) non-profit charitable organization."

It's more than just the right of first sale; with software that is licensed via server-side communication, nothing prevents the company from terminating your authorization for any reason, and you have basically no recourse at that point, other than to sue.

There's a lot wrong with software in the modern era.

Because they don't know where the small sources of waste are, and it takes time to fix them. If you need an immediate reduction right this second, the only thing you can do is surgical cuts, which means laying people off. Fixing the small sources of waste has to be an ongoing process that continues forever, and most of the interesting fixes actually cost *more* money in the short term to save money in the long term.

They might, if it happens to be at the end of the year when they do the cuts, and if that spending happens to be enough, but most of the time when this happens, they're looking for 30% cuts, not 1%. And finding thousands of fractional-percent cuts takes too long.

Not really, no. Inflation changes the value of the dollar. That means the government's debt also becomes less expensive every year, assuming all else is equal. And inflation causes increases in income, both for businesses and individuals, which means revenue should be increasing roughly proportionally. If it isn't, then that means the tax code is failing to properly capture percentages of actual gains, and this is something that needs to be fixed structurally.

In inflation-adjusted dollars, treasury revenue is going up, at least on average. From 2015 to 2025, tax revenue increased by 18.3%. Meanwhile, assuming Gemini isn't gaslighting me, the U.S. population increased by only about 6.6% in that time. So not only is revenue increasing after adjusting for inflation, it is also increasing relative to the population size after adjusting for inflation.

I can't tell you why service seems to always be getting worse. Maybe it is because we're spending rapidly increasing amounts of money on the most inefficient healthcare system in the first world, driven by a combination of lack of a public option or single payer system, poor auditing of payments, massively delayed payments that cause small healthcare providers to struggle to survive and force consolidation into giant regional monopolies, and probably a lot of other things that I don't know about because I don't work in that field.

When you end up having hyperinflation of your medical insurance costs, it eats a bigger and bigger piece of every other part of the budget. And the federal government is not immune to that.

There are probably other reasons as well. That's just the first one that comes to mind.

This is moot, because as you can see from the chart, inflation-adjusted revenue increased rather rapidly during that same period.

Maybe a percent or two, but with a budget is big enough, that's still a lot of money that could be used for something else.

The point is not that any of these things individually will result in big gains. The point is that there are a lot of different small inefficiencies that add up to a bigger inefficiency.

For example, for some reason, when the IRS sent out their findings for tax exempt status, a group that I work with never got the determination letter. And the IRS had no straightforward mechanism to resend the letter. Fixing it involved hundreds of phone calls before we reached a person who could help, and then waiting for someone to print it and mail it to us. All of this stuff should be in electronic records, and it should have been trivial for us to directly get a new copy electronically from their computer systems without requiring a person at the IRS to intervene.

Every time a person has to do something because a computer lacks code to do it, that is an example of government waste. It probably isn't worth fixing all of them, because sufficiently rare things could take decades to recover the cost of coding them, but that doesn't mean that someone shouldn't triage them, catalog them, prioritize them so that the scope is fully understood, because when you do that, you may find other people coming in later and saying, "If you do that, it will save me time on related task [x]," and that might then turn out to push it into "implement this ASAP" territory. Without documenting the state of things, those discoveries won't ever get made, and nothing will improve.

And the IRS has multiple incompatible login systems that use different credentials, multiple sites that expose different parts of the same access to information about your business/charity, etc. all of which have to be maintained, resulting in massive levels of redundancy, not to mention causing massive confusion for anyone who ever has to access them, wondering why it says they don't have an account even though they had to have one to fill out previous IRS paperwork. Replacing them with different views into the same data (with access right limits, presumably) in the same online system would likely save significant money, both in terms of software maintenance costs and server operation costs.

And how much auditor time could be saved if they trained AI models on previous audits and used that as a starting point for flagging suspicious returns and/or filtering suspicious returns flagged by existing automation? I don't have any idea, but I would not be surprised if that approach eventually produced meaningful long-term savings.

And every time they send out tax forms, what manual processes have to happen to distribute advance copies to companies like TurboTax, and how much time would be saved if we had a centralized, modern electronic version of all of the forms, rather than PDFs, with an open source implementation, complete with code to populate one form from another, etc.? Maybe it would cost more initially, but would save money in some other areas, like making it easier for auditors to recompute the taxes after fixing errors in data entry. I'm not sure, but these are the sorts of efficiency wins that should be looked at.

So in that one division alone, there are glaringly obvious inefficiencies that, if fixed, could result in considerable cost reduction. Similarly, every time you deal with someone at the Social Security Administration or (at the state level) the DMV and they tell you that the computer system is down and they'll try again in a minute, that's an example of government waste. It's a system that isn't working correctly, which as a result, wastes the time of hundreds, thousands, even tens of thousands of government workers on an ongoing basis.

There's no reason to believe any other part of the government is any better. Government IT is known for being disastrously slow at modernization, and it costs taxpayers a lot of money because our government doesn't spend the money to bring those systems up to date in a timely manner.

These are just some examples that are obvious from the outside looking in; there are probably many less obvious examples that would be obvious to someone who works there every day. And that's the point. The people at the top can't see what wastes the time of the people at the bottom, because they don't have visibility into their minute-by-minute activities (and even if they could, they would have a hard time filtering the flood of data into something useful). So you have to drive efficiency from the bottom up, and our government does not do this, so we can never really know whether that inefficiency adds up to half a percent or ten percent.

We can't get a complete picture without going to the people at the bottom of the org chart and asking them what could be done to make them more efficient, what could reduce waste, etc. It's a relatively easy low-hanging-fruit task, so we should do this. :-)

I hope that makes my position clearer.

"These things"? You mean the government shouldn't be drafting regulations for government, which is what we're talking about here? Instead, private industry should be telling the government what to do?

I don't care how much of an idiot you are, you're simply too stupid to respond to further. I don't want things to be as they were in my childhood. Back then, things were a mess. BECAUSE government tried either to micromanage everything or manage nothing at all. The idea of a third way, where governing is about just that, placing control mechanisms in place but not do the management, is obviously far beyond your pea-brain.