Hacker News new | past | comments | ask | show | jobs | submit | garspin's comments login

Things that have changed recently that may affect the trend -

faith in the stability of US govt

declining need for petro dollars

increase in US debt as %age of GDP

(IANA Economist)


Speaking as an non-american reading local & global news sources....

...does Trump not understand 2nd round effects ?

From afar he seems to be intent on Making America Great for Billionaires at the expense of any country less powerful than the US.


I already read article about how Trumps people are fuming at Zelensky as the cause of Vance and Trump hysterical meltdown

So, no, Trump does not understand that. He is protected from that.


> and then you found that it offered marginal-but-not-amazing improvement over the previous version.

Then call it GPT 4.1 and allow version space for the next iteration.

I think the label V4.5 is giving the impression of more than marginal improvements.


I use minbrowser.org/ Some sites disallow it... min suggests changing the user-agent setting to something like - Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/121.0


An alternative explaination...

Deepseek is a side project for a hedge fund.

Shorting NVIDIA & releasing everything including the source would have a high probability of being hugely profitable, with almost zero downside if it went unnoticed.


That would be an interesting 'insider trading' work-around: use inside knowledge of a private, non-US company to trade correlated stocks in public US companies.

If it was a public US company releasing Deepseek, and you shorted nvidia based on inside knowledge, presumably you'd fall foul of insider trading rules?


An interesting spin. But not an easy feat in any way. Yes, if you see the benchmark results, you can start speculate about its impact on the stock market

But it’s a consequence of your highly concentrated effort on the primary project (DeepSeek), not a side project.


Joseph Heller (of Catch-22 fame) was one of them.

His response to a billionaire was 'I have something you will never have - enough'.


perfect, just perfect!


LLMs are text compression algos. They are very good at memorising & retrieving text related to the user input. They can even pass bar exams based on that - some misinterpret that as intelligence.

However, no amount of scaling will change a text memorisation algo into a symbolic reasoning or composability algo, both of which are necessary for progress towards AGI.

So yes, LLMs are a dead end in the quest for AGI. However, they have their uses as a google/stackoverflow replacement.


I implemented it in the browser. For demo & src, see https://svelte.dev/playground/2c1bf42e0d2a4cebb38b907fa7f90a...

Adjust separation, cohesion and alignment manually, or use the presets.


> Pulling that off is the social skills test.

Surely the most significant side effect of any F2F interview is to evaluate social skills. An answer that the interviewer fully expects to be a lie just isn't required.

Anyway... "Why do you want to employ me over other candidates ? "


Who said anything about lies? In my experience, the primary goal of F2F interviews is to evaluate social skills like communication.


The only goal is to evaluate social skills.

For people who lack social skills the best way to pass those test is to lie, wich is pretty easy.


Google 'polarized training'. It's 80% really slow conversational pace jogging, and 20% really hard, struggle to breath flat out sprinting. Just avoid the intermediate jogging that's comfortable, but unproductive.


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: