prng2021

Leveled the playing field? Strides in relative wages? See the chart here under the section labeled "Inflation Adjusted Reality" and you'll see how incorrect your statements are:

https://www.advisorperspectives.com/dshort/updates/2025/01/0...

We still live in a country where federal minimum wage doesn't change for decades. You should also research the concept of subminimum wages and I think you'll be shocked.

If you already knew about all of the above and still think there was anything close to a leveling of the playing field, you're choosing to ignore facts.

What should they have said to make it non-misleading?

Sounds like you’re the one person in the world that has figured it all out. What are you waiting for?

Every reference to Amazon and Alexa in your comment would still be completely accurate if replaced with Google Home or Apple HomePod.

So the vision for Amazon is their box and underlying AI would forever be ahead of Google and Apple? So incredibly far ahead in fact, that Google and Apple’s massive amounts of user data and decades of habitual use (through Google search, Apple/Google maps, YouTube, Apple/Google pay, and a dozen other multi-billion dollar services) still isn’t enough to prevent Alexa’s AI domination? Again, it all seems like a massive waste of time.

How would a smarter Alexa allow them to “dominate AI”? What even is the vision here that would allow domination over competitors like Google and allow a massive boost in profits? It all seems like a giant waste of time for them.

"Deep research was trained using end-to-end reinforcement learning"

Does this mean they skipped supervised fine tuning like DeepSeek did with R1?

I think a more accurate description is they’re a standard company with a consulting/professional services arm.

Hilarious. This is from his wikipedia page:

“in 2019 a $1 billion lawsuit was filed against several Israel supporters, including Ellison. The lawsuit accused Ellison and others of conspiring to ethnically cleanse Palestinians from Israeli-occupied territories, committing war crimes, and funding genocide.”

From the article I linked:

“Our internal tests reveal a leap in accuracy from 98.97% to 99.56%, while customer test sets have shown an increase from 95.61% to 98.02%. In some cases where the document photos are unclear or poorly formatted, the accuracy could be improved by over 20% to 30%.”

It was never “solved” unless you can point me to OCR software that is 100% accurate. You can take 5 seconds to google “ocr with llm” and find tons of articles explaining how LLMs can enhance OCR. Here’s an example:

https://trustdecision.com/resources/blog/revolutionizing-ocr...

Determining whether the latest off the shelf LLMs are good enough should be straight forward because of this:

“Some participants have dedicated years of their lives to the program—like Alex Smith, a retiree from Pennsylvania. Over nine years, he transcribed more than 100,000 documents”

Have different LLMs transcribe those same documents and compare to see if the human or machine is or accurate and by how much.

I agree but do see 1 realistic solution to solve the problem you describe. Every product on the market is independently integrating a LLM right now that has access to their product’s silo of information. I can imagine a future where a corporate employee interacts with 1 central LLM that in turn understands the domain of expertise of all the other system-specific LLMs. Given that knowledge, the central one can orchestrate prompting and processing responses from the others.

We been using this pattern forever with traditional APIs but the huge hurdle is that the information in any system you integrate with is often both complex and messy. LLMs handle the hard work of handling ambiguity and variations.

I hadn't thought about that, but I'd actually like to see if it's better to do the opposite. Insert half a leap second twice as frequently as we currently do with a leap second. If that's successful, switch to inserting a quarter leap second and so on, maybe until we get to a state where we adjust a tiny fraction of a leap second on a daily basis. The point is that even if you have a system screw up time adjustment 1 day, it's such a small time error that shouldn't affect anything, and people can immediately work on fixing it.

And actually, what you explained about waiting until some significant threshold like 5 minutes is already something we do today in a way. It's just that we wait years and our threshold is 1 second before we actually make a change.

A market crash won’t be a silver bullet. I’ve known several gambling addicts and that’s not how addiction works. They don’t just stop because they lose a bunch of money. They always hold out hope they’ll be able to win everything back “next time”.

Most people will read the conclusions from this research and believe we’d all be better off without Walmart. The question really is whether you think we’d be better off getting rid of all big corporations and reverting to a much much smaller economy with only mom and pop shops. It’d be a world where businesses have no economies of scale. As a result, would you be ok with a country of significantly less consumption of everything and significantly fewer conveniences (ie free overnight shipping)? I mean electronics, food, clothing, and everything else you buy today sold by Amazon, Target, Walmart, etc. I don’t have a very strong stance either way on this, but that’s the reality you have to be ok with if you believe big corporations are bad.

“I would imagine that companies like Meta have such good automated moderation that what remains to be viewed by a human is practically a firehose of almost certainly disturbing shit.”

This doesn’t make sense to me. Their automated content moderation is so good that it’s unable to detect “almost certainly disturbing shit”? What kind of amazing automation only works with subtleties but not certainties?

I never said Facebook is the responsible party. I’m saying these workers deserve our sympathy and I’m saying it’s not a case of people who had a simple choice but willingly chose a job that caused them PTSD.

I don’t think Facebook is blameless though. They practically brag about their $40B of AI spend per year and absolutely brag about how advanced their AI is. You can’t focus some of your R&D to flag content that’s instantly recognizable as disgusting content, like pedophilia, necrophilia, and beastiality? There’s already a ton of pre-labeled data they can use from all these workers. No, they don’t get a pass on that. I think it’s shameful they focus all their AI compute and engineering on improving targeted ads and not put a major focus on solving this specific problem that’s directly hurting so many people.

You think people who took these jobs had a list of job offers and were jumping for joy to be able to pick this one out? Or that they stuck with it after the first 5 minutes of moderating necrophilia because they believed other jobs would have similar downsides? You’re so out of touch with the real world and hardships people face trying to make a living for themselves and their family.

I’m confused about the excitement. Are people just flat out ignoring the sentences below? I don’t see any breakthrough towards AGI here. I see a model doing great in another AI test but about to abysmally fail a variation of it that will come out soon. Also, aren’t these comparisons completely nonsense considering it’s o3 tuned vs other non-tuned?

> Note on "tuned": OpenAI shared they trained the o3 we tested on 75% of the Public Training set. They have not shared more details. We have not yet tested the ARC-untrained model to understand how much of the performance is due to ARC-AGI data.

> Furthermore, early data points suggest that the upcoming ARC-AGI-2 benchmark will still pose a significant challenge to o3, potentially reducing its score to under 30% even at high compute (while a smart human would still be able to score over 95% with no training).

Agreed but it’s not baffling. To me this is just another case of marketing disguised as research. An AI company whose sales pitch to differentiate themselves in the market is being hyperfocused on safe AI. So they participate in research that shows AI is “lying” and therefore can be dangerous. That’s why we should entrust Anthropic amongst all the AI companies! All these companies are run by people and they all have the same motives. Money and fame. Secret scratch pad of AI’s inner thoughts? Give me a break.

The author already mentions the benefits of streaming devices. What difference does it make whether you have a dumb or smart tv if the only functionality you use is from a streaming device?

This is useless. The SEC requires public companies to list their entire c-suite on their annual proxy statements which anyone can find with a 2 second google search.

What do you think the number of people and average amount they have invested is based on? I’m saying it’s based purely on emotion and speculation. It could be a sunny day so people feel good and put more money into Bitcoin. That’s quite different from egg prices skyrocketing after chickens get bird flu or stock price skyrocketing after an amazing earnings report or the dollar skyrocketing because of a fast growing economy.

$1k, $10k, $100k. Watching this nonsensical hype at every stage left me in disbelief. Funny thing is after all these years I’m finally realizing there’s no reason whatsoever why it couldn’t hit $1M, $1B, or go back to $1.

Anytime we see a dollar symbol we assume something has value like a traditional asset, tied to tangible or measurable things like material costs, a company’s quarterly earnings, etc. But cryptocurrencies are totally different. This thing’s dollar value per unit wildly fluctuates literally hourly, because it’s based purely on human emotion and speculation.

This sounds so useless that it could be an Onion article. What am I missing?

“There are no penalties for young people or parents who flout the rules.

Social media companies also won't be able to force users to provide government identification, including the Digital ID, to assess their age.”

The article says they used a forked version of Vitess with customizations. What were they and how did you address that when migrating?

I thought IPs are trivial for hackers to spoof. Given that and the fact that providers like AWS publicly publish their IP ranges, I don’t understand why IP filtering is so common.

Logging into SO using a Google account is just authenticating a user using OAuth and is a different topic. A typical scenario for using SCIM is you get hired at a company and as part of onboarding, the company’s Okta tenant sends a SCIM request to automatically create an account for you in their corporate Stack Overflow tenant. As part of your SO profile creation, Okta can include an optional Okta-specific identifier which is how Okta distinguishes you from everyone else. This is the external ID. Why’s this field useful? Well it’s one way that Okta can quickly find your account in SO in the future to make profile updates like updating a job title profile field.

I’m glad this author wrote this up. The guys in All-In really need to be called out for all their bs.

Here’s episode 191 Aug 9, 2024 where Chamath is talking about a Google breakup possibility:

“The big O outcome though is more if you go back to the Ma Bell kind of thing where the company gets broken up. I think that the odds of that are extremely unlikely. I think the big O outcome is probably something that you can pretty safely take off the table.”

And here’s episode 199 Oct 11, 2024 where the host rewrites history:

“We have an update on the DOJ's antitrust suit with Google. It looks like they're going for the break up as Chamath predicted”

What??

Lol. Your solution is to change a bunch of recorded money transactions? Oh shoot I messed up. Let me just change all these cash transactions temporarily to stripe. Oh shoot I also messed up my sales tax for these geographic regions. Let me just change all the dollar amounts temporarily. That’s hilarious.