HNHacker News
TopNewBestAskShowJobs

therobots927

1,029 karma · joined August 28, 2023

submissionscomments
therobots927··on Yann LeCun says Dario Amodei "knows nothing about AI effects on jobs"
That is complete bullshit. If you actually believe that, then you’re almost certainly either a manager or an IC specializing in CSS, HTML, or JavaScript.
therobots927··on Yann LeCun says Dario Amodei "knows nothing about AI effects on jobs"
Lying to hype your product is not a “meta game”. It’s a short term strategy where you trade your future relevance and credibility for a swimming pool full of cash.

Yann LeCun is a Turing award winning research scientist who will be remembered as a scientist and a visionary.

Amodei’s name will go the way of WeWork’s Adam Neumann - it will be mostly forgotten, and only remembered on occasion as a massive and embarrassing fraud.

therobots927··on Yann LeCun says Dario Amodei "knows nothing about AI effects on jobs"
Yann is a serious person in a very unserious field. He will be vindicated sooner than you think.
therobots927··on The electromechanical angle computer inside the B-52 bomber's star tracker
I’m sorry. I’m in a bad mood and that was unecessary. That being said, given the current hyper militarized climate in Silicon Valley, I find this detachment of the science / engineering from its use cases to be more than a little distasteful.
therobots927··on The electromechanical angle computer inside the B-52 bomber's star tracker
I’m about to read King Leopold’s Ghost. Great choice in username.

I must say it’s a little disappointing that things like “secret bombing campaigns” getting declassified don’t lead to much public response.

therobots927··on The electromechanical angle computer inside the B-52 bomber's star tracker
And some people, specifically Vietnamese and Cambodian civilians, were on the receiving end of your fun little brain teaser.

And other people, like Henry Kissinger, drew random dots on a map to tell it where to drop the bombs. https://en.wikipedia.org/wiki/Operation_Menu

therobots927··on Opus 4.7 to 4.6 Inflation is ~45%
I absolutely do not accept that argument. It’s clear models hit a plateau roughly a year ago and all incremental improvements come at an increasingly higher cost.

And junior devs have never added much value. The first two years of any engineer’s career is essentially an apprenticeship. There’s no value add from have a perpetually junior “employee”.

therobots927··on Opus 4.7 to 4.6 Inflation is ~45%
Wow this is pretty spectacular. And with the losses anthro and OAI are running, don’t expect this trend to change. You will get incremental output improvements for a dramatically more expensive subscription plan.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
Google is doing something != they are doing the rational thing.

Suggesting otherwise is defintionally argument from authority.

You know you can just admit you’re a sophist and we can move on.

therobots927··on Hyperscalers have already outspent most famous US megaprojects
That’s fine and dandy but it won’t do anything for most people with serious health problems.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
Argument from authority with a little whataboutism thrown in. I’m done wasting time with you.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
What exactly does “personalized medical treatment” entail?

Writing prescriptions?

Ok, I can see how AI could theoretically do that (assuming it doesn’t hallucinate and kill a bunch of people). Oh and don’t think it’ll be so easy to give AI the legal authority to prescribe controlled substances. And insurance companies may take issue with expensive prescriptions written by a chat bot.

Perform surgeries? Stitch wounds?

That’s decades away. And that also opens a legal can of worms. Maybe the AI lawyers can figure something out.

therobots927··on Hyperscalers have already outspent most famous US megaprojects
How much has china invested in GPU buildout?
therobots927··on Hyperscalers have already outspent most famous US megaprojects
They aren’t building out datacenters the way the US is. The arms race is a myth.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
A lot of this is committed capital and the datacenters haven’t even broken ground.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
Well it can build a killer front end and that’s great at impressing investors.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
But the AI fanatics claimed that AI would solve cold fusion, making that whole thing moot.

The only problem is, if AI doesn’t solve cold fusion, we’re back to square one. And a few trillion dollars in the hole.

therobots927··on Hyperscalers have already outspent most famous US megaprojects
Same. He’s very knowledgeable about this and very skeptical. Not to mention hilarious.

I’m getting my popcorn ready for the bubble pop.

therobots927··on Hyperscalers have already outspent most famous US megaprojects
So your argument basically boils down to, the datacenter build out is not a waste of resources because if it was, these companies wouldn’t be building them.

Got it.

therobots927··on Measuring Claude 4.7's tokenizer costs
Given that they haven’t even gone public yet, doesn’t that seem like putting the cart before the horse a bit? And if they’re already enshittifying it won’t be long until the other placers start doing so as well. Have we passed peak LLM intelligence and are we now watching it degrade as they fail to roll these new advanced models out to their increasing user base? Are the finances not adding up?

Lots of questions.

therobots927··on Measuring Claude 4.7's tokenizer costs
It’s nice to see comments like this. It makes me feel less crazy. Something very weird is going on behind the scenes at Anthropic.
therobots927··on Claude Opus 4.7 costs 20–30% more per session
OpenAI is dealing with exactly the same energetic and financial constraints as Anthropic. That will become apparent soon.
therobots927··on Hyperscalers have already outspent most famous US megaprojects
What % of GPUs are running self driving software or robotics?

And what is the ROI on either of those right now?

therobots927··on Hyperscalers have already outspent most famous US megaprojects
The problem is that once built, railroads provided economic value right off the bat.

I would love to hear about the economic value being generated by these LLMs. I think a couple years is enough time for us to start putting some actual numbers to the value provided.

therobots927··on Claude Opus 4.7 costs 20–30% more per session
As a regular listener of Ed Zitron this comes as absolutely no surprise. Once you understand the levels of obfuscation available to anthro / OAI you will realize that they have almost certainly hit a model plateau ~1 year ago. All benchmark improvements since have come at a high compute cost. And the model used when evaluating said benchmarks is not the same model you get with your subscription.

This is already becoming apparent as users are seeing quality degrade which implies that anthropic is dropping performance across the board to minimize financial losses.

therobots927··on Measuring Claude 4.7's tokenizer costs
That’s the joy of purchasing an intangible and non-deterministic product. The profit margin is completely within the vendor’s control and quality is hard for users to measure.
therobots927··on Claude Opus 4.7
The only way it’s more affordable is if anthropic burns cash to keep their corporate clients.
therobots927··on Claude Opus 4.7
Who cares about a new model you can’t even use?
therobots927··on Claude Opus 4.7
Here’s the problem. The distribution of query difficulty / task complexity is probably heavily right-skewed which drives up the average cost dramatically. The logical thing for anthropic to do, in order to keep costs under control, is to throttle high-cost queries. Claude can only approximate the true token cost of a given query prior to execution. That means anything near the top percentile will need to get throttled as well.

By definition this means that you’re going to get subpar results for difficult queries. Anything too complicated will get a lightweight model response to save on capacity. Or an outright refusal which is also becoming more common.

New models are meaningless in this context because by definition the most impressive examples from the marketing material will not be consistently reproducible by users. The more users who try to get these fantastically complex outputs the more those outputs get throttled.

therobots927··on Atlassian defends firing engineer for suggesting CEO is 'rich jerk'
Well, is he a rich jerk?
← PreviousPage 8 of 28Next →