They are both great and also biased. Because LLM training data includes vast libraries of history books, including such works as Dark Alliance, it is capable of running down paths that few historians would walk. I recently used an LLM to find an unimpeachable source for the CIA’s connection to cocaine trafficking aircraft. This was not available through a simple Google search because the site was likely deindexed.

They are promoting the disproven conspiracy theory that illegals are impersonating US citizens and then casting ballots under false names.

Greg Palast’s reporting proved that they are trying to purge minority and left leaning voters from voting rolls.

I know what you mean. All the code generated with ai was loaded with problems. Specifically it kept forcing my api keys into the code without using environmental variables. But for basic coding concepts it has so far been perfect. even a 3b model seemingly generates great definitions

Snippets are a great use.

I use StableCode on my phone as a programming tutor for learning Python. It is outstanding in both speed and in accuracy for this task. I have it generate definitions which I copy and paste into Anki the flashcard app. Whenever I’m on a bus or airplane I just start studying. Wish that it could also quiz me interactively.

It’s a huge disaster in the making. There’s probably a game theory model that explains what’s happening but damned if i know it. AI is a trap.

At the beginning of LLMs as a consumer tech phase, the big tech companies believed that they would soon generate synthetic data without human labor costs. But synthetic data caused model collapse when used as training data. So they still need humans.

But now their competitors are using scrapers and summarizers and those are expensive so they have to extract more value from search traffic and that means fewer referrals to sites. Fewer referrals to sites reduces original content and leadabto less traffic and training data for their models.

The only two ways out are to create an international regulatory body ornto break up facebook, microspft, and google.

I appreciate your informed response but no system other than advertising-abstinence is fool proof.

Im saying this as a supporter. My browser of choice is firefox and I send them money regularly. And I understand their need to generate more revenue. But there has never been a company who has sold customer data discretely. My understanding is that every piece of data that’s sold can be de anonymized when combined with other data sets. And the data is horsetraded until it gets into some very marginal actors’ hands.

Mozilla’s need for money is largely driven by massive mismanagement. It should have been fully funded in perpetuity through establishing a foundation that operates off interest payments but they decided to try and build a headquarters in Mountainview. They also operate offices in some of the most expensive cities in the world. They have made expensive software aquisitions. These are not necessary and have only whetted mozilla’s thirst for other revenue sources. It’s guaranteed that they will look for more customer data to sell because that’s the path of least resistance.

I wish them luck but I also wish they’d not chase advertising money.

Even if Mozilla takes precautions to avoid de-anonymizing our data, any private data sold to data brokers becomes a part of the puzzle for learning our identities

https://en.wikipedia.org/wiki/Data_re-identification

Even knowing something a trivial as two movie ratings led to a 68% success rate in learning an identity.

Everything you wrote lined up with the article on wikipedia so if you got something wrong I didn’t see it.

I’m referring to the book “This Time Is Different: Eight Centuries of Financial Folly” the title of which mocks the oft repeated defense of bubble investors:

https://www.nber.org/system/files/working_papers/w13882/w13882.pdf

But their point is that every single asset bubble ended up popping, despite the protections instituted by banks and governments. They also point out that the bubbles have been getting bigger and bigger

I’m not sure what you mean, but no, I don’t think that and I didn’t write that but i can understand the confusion because it’s not well known how QE works. Some forms of QE prevent crashes. The Fed can achieve this by taking the bank’s failing debt instrument off the books, and swapping it for a t bill.

Sorry I appreciate your comment. So I read (erroneously?) that central bankers had done away with the reserve ratio in the fractional reserve banking article. And that just seems like a reckless thing to do given how prone to bubbles our economy is.

One of the main points in “this time is different” is that despite the math, we are experiencing greater and greater asset bubbles and at no point in world history were things actually different.

Thanks for the reply. I hope you don’t let my spelling or use of ex nihilo (this is the exact language used by the fed and economists, I didn’t just make it up) turn you off, because at a policy level they are pursuing policies that keep real estate prices high.

This seems like an already failed banking model which places lenders at the front of the pack and will lead to only larger asset bubbles. Japan’s Kiretsu system of banking led to banks taking out loans to cover up their own investment losses as they had put their money into an asset bubble which collapsed. Banks then committed wholesale fraud by disguising such losses on their books. The Japanese government then used quantitative easing. They create money ex nihilo, swap the money for a t bill, then they bought the toxic assets by giving t bills to the bank. The bank doesn’t sell the t bill, they merely collect interest on it.

The main effect is a system in which bubbles are never popped and consumers suffer a declining standard of living in order to keep asset prices high.