randcraw

And it had a MUCH bigger budget than the original, as well as the assurance that it wouldn't be cancelled at a moment's notice.

Frankly, I think we very much _do_ need to be reminded that LLMs are in fact _not_ eating the world.

It's at the forebrain of every executive in America that LLMs imbued with RLHF are on the cusp of replacing every job they can imagine, from robot chauffeurs to carpenters to surgeons. But as this paper suggests, the integration of generally capable vision systems with basic tool handling that will be needed to replace the specific skills essential to most human jobs -- especially using only today's frontier models -- still takes us unforeseeably and undeniably far into the realm of sheer fantasy.

I would have thought that the confidence scores available from CNNs but not from LLMs _would_ be valuable.

Clinically, I'd imagine that choosing the right final diagnosis with relatively low certainty is often less useful than reporting the level of confidence (high or low) of each candidate diagnosis. A set of plausibles is surely better than a single wrong ID.

Clinical tests, especially when used by nonspecialists, are useful not only as diagnostic endpoints. Often they're essential to separate plausible diagnoses from implausible. If a CNN can rule out all outcomes but one with high certainty, that's nothing to sneeze at.

And if it could also _explain_ why the certainty of each candidate was high (or low) -- as non-LLMs often can do better than LLMs. That should have value too, especially in establishing an audit trail to defend the chosen diagnosis.

I think Thompson has no understanding of endeavors like pure math or the theoretical sciences. In the words of Einstein, "I want to know how God created this world. I am not interested in this or that phenomenon, in the spectrum of this or that element. I want to know his thoughts; the rest [is] just details."

I think mathematicians aren't bemoaning their lost sunk costs. They're now aware that they may never again have the joy of being the first to pull up the edge of the tarp of ignorance and see the raw truth beneath _for the first time_. Their voyages of discovery, which give life meaning, may never leave dry dock again.

Your loved ones have replaced you with more agreeable robots. Now no one cares whether you live or die.

Still disgusted?

I'm reading Paul Theroux's "True North", on his recent tour of Canada, the country of his birth. Now 84, he claims it will be his last travel book.

Just finished Jill Lepore's "The Rise and Fall of the Artificial State" -- a history of the popular fascination (esp. among idealistic tech types) for ending human-based governance and replacing it with computer automation (esp. via AI). I appreciated her coverage of the rise more than her case for the fall.

IMHO, audiobooks <--> podcasts. The act of reading is unique, unlike any other medium.

Don't agree. If you know no facts, you'll be unable to think even the most basic thoughts. The more you know, the deeper your thinking can be as well as the more subtle.

For every fact you learn there is a necessary context and numerous exceptions when it's applied in other contexts. If you expect AI to spit all relevant facts at you every time you're clueless, you will have no idea when AI is wrong or lying. You will have to trust everything AI (or its owner) tells you.

I get reCaptchaed-to-death all the time in iOS and MacOS using both Firefox and Brave. I use a big name VPN which probably makes it worse, since I'm routinely blocked outright by Cloudfare services, assuming cluelessly that I'm a bot.

It turns out that some foods counteract the formation of kidney stones. If you drink milk when you eat oxalates, the oxalates will not fuse into stones and will be excreted normally in your urine.

Oxalates are present in tea too, though not coffee, IIRC.

And there were an unimaginable number of tax loopholes available back then that ultimately reduced the individual tax rate enormously. You just had to reinvest your income in a multitude of increasingly bizarre tax dodges devised by increasingly byzantine tax laws.

I think reading short stories is a great way to develop both writing and storytelling skills. A SS compresses the storyline and character development into a more palatable and succinct morsel than can a full-sized book. And the brevity of SSs should make them more digestible.

Classic literature was written in an era when readers had more patience and fewer distractions than today. But classic short stories should offer the best of those, IMO, better-written times: more thoughtful exposition in fewer words.

AI-cloned shows will suffer on day #1 what the original show suffered after the 5th year or so -- the plausible and original plot lines have been used up. So the AI-gen story lines will be implausible or derivative copies of familiar plots from the original series or other shows.

"The Machine-Made Muse" https://www.youtube.com/watch?v=kM4xISBsUcY

Only Data looks and sounds fully 'natural' in this 10 minute clip of STTNG, but it nicely demonstrates the imminent plausibility of synthetic TV / film. And its affordability.

Feed it a handful of short stories and you can automate the creation of an entire TV series.

> Points to the future where my bots talk to your bots. I think that’s healthy.

I can't imagine why. You have become a robot. You've automated yourself out of existence. And due to your consistently mindless replies, everyone now knows there's no point in talking to you ever again.

How is that healthy?

Agreed. When tools evolve significantly the old production process has to be reworked into a new process and that takes time. In complex production lines that will include changes to subcontracting, supply chains, and even finance models. The changes arising in software production today will take years to stabilize into a new normal, even longer if the tools continue to advance in capability as LLMs are doing. And many software-dependent companies will fail in that time for moving either too fast, too slow, or most likely, too short-sightedly.

Or you could just submit a photo of someone else's poodle.

In the late 1960s daily TV news coverage revealed the horrors of war to the American public as never before (just as Brady's photos had done in the Civil War). That palpable gruesomeness was instrumental in the rising tide of unpopularity enjoyed by the war (and Johnson) by the elections of 1968.

Emergency is the mother of making exceptions. During both US gulf wars, the army broke from its long tradition of over-regulating new systems before they could be fielded, especially for small semi-autonomous platforms like flying drones and robots. Entirely new systems as well as major updates to fielded system were routinely prototyped in-country / on-the-battlefield in Iraq and Afghanistan. The demand for new capabilities and fixes was simply too great not to ship ASAP.

I look forward to following this issue. We're at an inflection point in AI adoption where, for some uses, it _must_ be trustworthy. Failure to stick to the facts must doom it for such uses until its wild hair reliably and accountably can be tamed.

I think you would most certainly NOT prefer any Constitution, nor especially any Bill of Rights, that was rewritten by today's version of "We the People".

> It’s crazy how much worse Google search results are now than they used to be.

I could not agree more. Each time I do a google search, I feel like I've stepped into a severe reality distortion field. The results are simply WORTHLESS. It's not just that the AI summary at the top answers the wrong question, but the sequence of hits thereafter wander off into the weeds almost immediately. I routinely have to constrain the search in multiple ways (time bounds, specific word rejection, etc) just to get one or two relevant hits.

Frankly, I think I'm going to have to MOVE AWAY FROM GOOGLE to a subscription-based search service just to get any useful search results again.

Present generations see the primary role for AI as being identical to the biggest disruption to US business in the past century — namely the gigantic sucking sound from the outsourcing of millions of US jobs in order to reduce labor costs. Of course they're terrified of it. By singing its praises AND IGNORING THAT CONCERN, Schmidt showed that he couldn't be more out of touch with the top priority of this generation — WILL AI TAKE MY JOB?

It's like Steve Forbes cluelessly joking that rise of Private Equity will make it easier for the unemployed to buy another racehorse. WTF?

I've noticed the same thing. IMO, the heyday of computing coincided with the peak of WiReD Magazine. Around 1995 interconnected computers and the WWW bootstrapped a revolution in spreading+sharing information that seemed certain to reshape cultural norms, organically, bottom up, for the better. It seemed the world would never be the same. Heady times. That sense of optimism lasted until the computer unicorns morphed into monopolies. After that the roles reversed and all us net denizens became unicorn fodder. It's hard to stay upbeat after you realize you're no longer in charge and are constantly being manipulated wherever you go online.

> Surely in the long run…

And that's the rub. PE is all about short term ROI at any price. Their business model doesn't take product superiority or brand loyalty into account. If a widget can be made cheaper, you do it, damn the collateral damage.

Yeah, I've seen the same thing happen to data miners in the pharma industry. An increasing fraction of young biologists have skill in basic statistical DM as well as web search proficiency sufficient to gather DM code analysis examples, even without using AI. In the very near future I expect almost all R&D exploratory DM will be done by pharma domain experts (biologists and chemists) rather than served by DM experts (computer scientists or engineers).

Yeah, IEEE Spectrum has responded to the dissimilar roles in SW dev by ranking programming language popularity contextually, by separating the project domains and ranking the languages only within each domain. That's a lot more useful than allowing the single dominant project domain to silence the recessive ones, as TIOBE does.

Rather than coining a new word like adaption, I'd call this acculturation. It's reshaping not only SW dev but natural language too -- how we read and write and how we speak.

Everyone knows that AI-written slop isn't worth actually reading. So when reading mass media content we skim over each paragraph's opening phrases rather than read it deliberately, sentence by sentence. We also do this while writing notes, dropping determiners, acronymming common phrases, and making references to characters/scenes in popular media. Now with the rise of vocal interfaces and ever shorter rounds of engagement, all this abbreviating will only exponentiate.

Yeah, 95% of the available advancement in computing is in people management, not technical mastery. Businesses much prefer to hire externally to serve any non-core capabilities, especially to minimize internal culpability should anything go wrong. That leaves little opportunity to think outside the box technically.

Clothes are a good example of what ails online shopping. When you physically visit a clothing store, you chose it knowing the quality and style of that merchant -- you thereby filter out a huge fraction of the market that you want to exclude from your search.

But online (because the available search criteria are so imprecise) your search brings up every possible form of clothing, especially the stuff that's a commodity (or hyped by major e-merchants) -- cheap, popular with 25 year olds, colorless, largely disposable. It's hopeless unless you yourself are a commodity -- indiscriminating, predictable, and totally average.