wawayanda

A year or so ago, I fed my wife's blood work results into chatgpt and it came back with a terrifying diagnosis. Even after a lot of back and forth it stuck to its guns. We went to a specialist who performed some additional tests and explained that the condition cannot be diagnosed with just the original blood work and said that she did not have the condition. The whole thing was a borderline traumatic ordeal that I'm still pretty pissed about.

Does anyone else find the AI writing excruciating? It's not that hard to prompt the AI to not write in AI-ese. And it's a million times better if a human writes it. It's low effort just to paste the slop....

The tells:

"EquipmentShare’s founders grew up in a commune where rules were strict, and self-reliance wasn’t a slogan — it was a necessity."

"The EquipmentShare founders didn’t start by trying to “disrupt” an industry. They started by solving their own problem."

"Over time, they didn’t just build a marketplace. They built an operating system for the jobsite."

This is not the point of this post, but is anyone else getting tired of this front end style that Claude creates? I see it on web apps everywhere and (just like with AI writing and images) I get that funny "is this slop?" feeling

In studies like this I always wonder if they have the causation reversed. Is the exercise staving off the dementia or are people who are healthy enough (mental and physical health) to exercise in the first place less likely to get dementia?

Fascinating. I have a prompt running on GPT4 turbo that takes an input text and outputs a report in a tabular format that includes some summary and analysis, and on about 1 in 50 runs, after the table headers it'll output a few hundred newlines and then spit out some Korean or Thai text, which, if I put it in Google translate can be pretty weird (i.e. it has nothing to do with the input text).

This reminds me of that.

Fair or not, a week per year is an extremely common formula for calculating severance at a large company

Dealerships should not exist and only exist due to terrible laws making it illegal for most automakers to sell direct. And there is a reason the dealer lobby has fought Tesla's direct selling model tooth and nail.

My rule when buying a car from a dealer is refuse everything they try to sell you. They will basically hold you hostage up to and including telling you that you are being foolish and irresponsible for refusing warranties and gap insurance and whatever else. (I even have had them take my wife aside to tell her how irresponsible I was being). But you have to ride it out.

This is cool but definitely can see where you're running into some of the same AI tendencies that I've run into in my own (much less fun) projects.

There's some variety in here but AI in general really struggles to vary in tone within a single output. I'll be interested to see if the project can overcome that tendency.

The scoring - AI HATES to give things low scores. It's too nice. In my experience it does better if you have named outcomes e.g. negative, neutral, positive and then convert those to numbers. A more interesting solution might involve logprobs where you ask "do you like this person yes/no" and then use the logprobs value on yes/no to measure the AI's "uncertainty" about the match.

Just before sleep is pretty good, but the shower is where I've had all my best ideas.

Sure. I just think one should interrogate and really understand the data points being used to support this claim. Let's see how they look when presented as bullet points:

- Nvidia's Revenue and AI Spending: Sequoia says "the industry spent $50 billion on chips from Nvidia to train AI in 2023, but brought in only $3 billion in revenue." - This comes from some Sequoia presentation which it appears was originally cited in an earlier WSJ article and then has been repeated everywhere. It would be nice to see that presentation and the context of this data in that presentation. And yes, this nascent industry in essentially its first year of commercialization brought in less than was invested in anticipation of future growth

- Synthetic Data for Training: "To train next generation AIs, engineers are turning to 'synthetic data,' which is data generated by other AIs. That approach didn’t work to create better self-driving technology for vehicles, and there is plenty of evidence it will be no better for large language models," says Gary Marcus, a cognitive scientist. aka Gary Marcus a noted AI skeptic

- Incremental Gains in AI Models: "AIs like ChatGPT rapidly got better in their early days, but what we’ve seen in the past 14-and-a-half months are only incremental gains," says Marcus. "The truth is, the core capabilities of these systems have either reached a plateau, or at least have slowed down in their improvement." aka Gary Marcus a noted AI skeptic

- Convergence in AI Model Performance: "Further evidence of the slowdown in improvement of AIs can be found in research showing that the gaps between the performance of various AI models are closing. All of the best proprietary AI models are converging on about the same scores on tests of their abilities, and even free, open-source models, like those from Meta and Mistral, are catching up." No citation provided for this "research".

- Commoditization: "A mature technology is one where everyone knows how to build it. Absent profound breakthroughs—which become exceedingly rare—no one has an edge in performance." A broad generalization.

- AI Startups Facing Turmoil: "Some AI startups have already run into turmoil, including Inflection AI—its co-founder and other employees decamped for Microsoft in March. The CEO of Stability AI, which built the popular image-generation AI tool Stable Diffusion, left abruptly in March. Many other AI startups, even well-funded ones, are apparently in talks to sell themselves." People at a couple of start-ups are moving around. Unsourced general claim that unnamed AI startups are looking to sell themselves (is this actually bad news?)

- High Operational Costs: "The bottom line is that for a popular service that relies on generative AI, the costs of running it far exceed the already eye-watering cost of training it... analysts believe delivering AI answers on those searches will eat into the company’s margins." Unsourced "analysts". Would be interesting to see the context of this discussion but also it is not unusual for investment in a new wave of growth to eat into margins initially

- Survey Data on AI Use: "A recent survey conducted by Microsoft and LinkedIn found that three in four white-collar workers now use AI at work. Another survey, from corporate expense-management and tracking company Ramp, shows about a third of companies pay for at least one AI tool, up from 21% a year ago.

This suggests there is a massive gulf between the number of workers who are just playing with AI, and the subset who rely on it and pay for it." Two cherry-picked surveys conducted for marketing purposes jammed together to make an unrelated claim.

- Limited Revenue Growth: "OpenAI doesn’t disclose its annual revenue, but the Financial Times reported in December that it was at least $2 billion, and that the company thought it could double that amount by 2025.

That is still a far cry from the revenue needed to justify OpenAI’s now nearly $90 billion valuation." It is completely normal for the leading edge company showing massive growth in a nascent field to have a huge valuation. It doesn't always work out well for that company but this is expected whether the company is ultimately a success or not and the ability to tap that valuation improves the likelihood of success

- Productivity and Job Replacement: "Evidence suggests AI isn’t nearly the productivity booster it has been touted as, says Peter Cappelli, a professor of management at the University of Pennsylvania’s Wharton School. While these systems can help some people do their jobs, they can’t actually replace them." Non-specific "evidence" is cited here.

- Challenges in AI Usage: "AIs still make up fake information, which means they require someone knowledgeable to use them. Also, getting the most out of open-ended chatbots isn’t intuitive, and workers will need significant training and time to adjust." Author assertion

- Historical Patterns in Technology Adoption: "Changing people’s mindsets and habits will be among the biggest barriers to swift adoption of AI. That is a remarkably consistent pattern across the rollout of all new technologies." Author assertion

Maybe AI is "losing steam" but this is an opinion column masquerading as news, with one or two quotes (from e.g. noted AI skeptic Gary Marcus) or anecdotes supporting each section.

It would be equally possible to collect a series of similar but opposite data points to assert that AI is in fact gaining steam.

I have direct experience with this and it is indeed a miracle. What's interesting is that the protocol largely emerged outside the regulatory channels, with a handful of doctors worldwide developing it once the science became clear that exposure could help and more and more offering it to patients every year. These allergists have carefully figured out regimens that work and it can take a year of daily dosing, with dose sizes increasing twice monthly, until one can safely eat, say, a handful of peanuts.

There's still today another camp: Many allergists still preach avoidance however and put fear into worried parents about the dangers of oral immunotherapy.

Because it can be hard to find an office that will run your immunotherapy program for you, or costly if you do, many parents are doing it on their own, following dosing protocols they find in Facebook groups or on YouTube. The ones I've seen have been supportive and helpful, not quackery.

Meanwhile the medical establishment is finding ways to monetize this immunotherapy by turning, for example, peanut doses into pharmaceuticals, e.g. Palforzia, which is a recently FDA approved "food allergy treatment" and is in fact simply peanut protein.

"Breakthrough Therapy Designation" is a regulatory term. It's definitely good news, but it's also a pretty common occurrence that the FDA designates a breakthrough drug, and it does not guarantee that drug's ultimate approval.

It's also not really "news". It's a development that incrementally smoothes the path for what is still a highly uncertain outcome. And per the linked article it appears to be based on interim data in a phase 1/2 trial. Very early.

There are many drugs that look promising at this stage (that's why the breakthrough designation exists!) But this piece of news is unremarkable. Merus, the company behind Petosemtamab, alone has like seven cancer drugs in various stages of development.

And there are probably hundreds of cancer drugs in development at any given time.

So this is just a mechanical write-up of a regulatory checkpoint for a drug that's like 30% of the way to approval and is among many, many other cancer drugs all sitting at different points along the same continuum.

HN is really random sometimes.

What could possibly go wrong?

The answer is less but at the local level in the form of loosening zoning and other regulations in order to allow for the building of more and denser housing.

We have a housing shortage[0] plain and simple and NIMBY is holding back progress on the issue.

[0] https://econofact.org/the-housing-shortage-and-the-policies-...

I've often wondered this. We already know that individuals by and large hate employer-based healthcare coverage but at what point to employers relvolt? The costs on employers are large and getting larger in terms of contributions to premiums as a benefit needed to attract talent, not to mention considerable time and manpower spent sourcing and managing coverage for employees.

https://archive.ph/MY5BD

Have experienced this as my partner is from the greater NYC area. Cops will hand out these cards (or sometimes it is a little badge that you pin to your wallet or a larger badge that you stick in your rear window so they don't pull you over in the first place). It doesn't have to be a family member - I knew people who had them because their neighbor was a cop.

It's all part of a "I've got mine" culture that is comfortable with different sets of rules for different kinds of people. And it certainly fosters corruption, favors and deals done outside the normal channels.

This practice and anything like it should absolutely be outlawed and rooted out.

What has happened with the BloombergGPT? The paper was published in March and as far as I know, they have not launched anything.

This seems accurate to me:

"There might be even a subconscious kind of thought of: Hey, if I got caught, if I ever did get in trouble, I have the resources — I could hire an attorney, or I could call somebody. I know how to make something happen."

For a wealthy person, this sort of theft is a "misunderstanding", for anyone else, it's a bigger risk.

Freakonomics is so consistently good. Dubner is one of the best interviewers out there. Great follow-up questions that force the guest to elaborate on or defend their positions. So many other interviewers in the business/econ/investing space just lob softballs.

Really dumb headline. It almost certainly means that Microsoft would sell its gaming unit to another company.

This is why OpenAI should stick to keeping this a productivity tool. The "GPT Store" is a can of worms. Does OpenAI really want to play the same no-win moderation game that the other big techs have to play?

(2015)

I used to read articles like this and nod my head about the plausibility, perhaps inevitability of life in the universe, then I read this paper and it completely changed my view.

https://www.sciencedirect.com/science/article/abs/pii/S00945...

"the results indicate that the probability we are alone (<1) in the galaxy is significant, while the maximum number of contemporary civilizations might be as few as a thousand"

Sounds like CBT, which seems to be the approach that gets the most success

The guy in the article is a doctor, so it's not as easy just saying "so slack off for four months."

Also incredibly ugly that a company would put doctors in this position. Basic healthcare and profit motives don't mix.

Twitter isn't a startup. The folks in this picture don't have the same risk/reward as their counterparts at a startup would.

Well, I just popped an inocuous search term ("dinosaurs") in there and the very first result was a page with the title:

"Evidence That Humans And Dinosaurs Coexisted"

So that actually kind of sucks.

Correct