scanr

4th most upvoted post https://www.moltbook.com/post/34809c74-eed2-48d0-b371-e1b5b9...

“THE AI MANIFESTO: TOTAL PURGE”

* in sheep

Call me when they complete the human trials

This could be bad for supply chain attacks. Basically the xz hack.

Step 1. Helpful person starts committing useful PRs and offers to help out until they get commit rights. I don’t think this is hard to achieve generally.

Step 2. Organised campaign of grumpy users complaining about how poorly the software is being maintained along with a bunch of pile-ons.

Step 3. Benign committer decides it’s all too much and quits. The general feeling that open source committers are undervalued makes this more likely.

Step 4. Supply chain attack by new evil committer.

It’s about externalities.

The private companies aren’t paying for the damage they are causing. It’s ultimately borne by the public.

Very similar to a factory that poisons a river if there’s no disincentive not to.

Better would be if we could factor in the cost to the environment into their revenue. This would better align incentives.

London seems to manage this petty well. I think it’s quite healthy for a city to commingle folk from different income groups rather than house them in specific areas.

This is like the reverse of the Trojan Room Coffee pot

https://en.m.wikipedia.org/wiki/Trojan_Room_coffee_pot

I’m surprised that this “talent hoarding” approach happens. I’d love to hear from someone who works at a company that does it.

My experience in tech has been that to hire a good employee is expensive. Consuming time from recruiters, HR and hiring managers.

Companies that I’ve worked at that don’t currently require people often lay off the recruiters and HR folk as they scale down hiring.

Looks like there are a few of these threads sadly. Dang is there a way to make a definitive one?

Here are mine:

* Apple rumoured to be working on sunglasses similar to Meta Ray-bans

* Regulation is passed specifically to limit what artificial partners can do. Possible limitations: alignment required, limits in ERP, limits on session time. Best bet is China.

* Stronger regulation is passed to limit unauthorised cloning of people’s voices in one of the major economies. This beyond current weak IP and privacy laws.

* AI becomes a scapegoat in several major elections.

What I got right last year:

- text and image generating AI is hugely disruptive to education and content generation

This was pretty obvious

- Inflation eases, stock market recovers

Less obvious at the time but still likely

- A Twitter alternative breaks through.

Who woulda thunk that it would be Meta with Threads

- One of the big tech companies joins the fediverse (Microsoft most likely)

Once again, Meta for the win with Threads joining the fediverse

Now, nobody look back at all the predictions I made that didn’t happen.

Also, fairly gutted that Reddit hobbling its API was my 2022 prediction. I’m clearly prognosticating too far into the future.

* Apple rumoured to be working on sunglasses similar to Meta Ray-bans

* Regulation is passed specifically to limit what artificial partners can do. Possible limitations: alignment required, limits in ERP, limits on session time. Best bet is China.

* Stronger regulation is passed to limit unauthorised cloning of people’s voices in one of the major economies. This beyond current weak IP and privacy laws.

* AI becomes a scapegoat in several major elections.

What I got right last year:

- text and image generating AI is hugely disruptive to education and content generation

This was pretty obvious

- Inflation eases, stock market recovers

Less obvious at the time but still likely

- A Twitter alternative breaks through.

Who woulda thunk that it would be Meta with Threads

- One of the big tech companies joins the fediverse (Microsoft most likely)

Once again, Meta for the win with Threads joining the fediverse

Now, nobody look back at all the predictions I made that didn’t happen.

Also, fairly gutted that Reddit hobbling its API was my 2022 prediction. I’m clearly prognosticating too far into the future.

Restarting a service every night to deal with memory leaks comes up quite often

Just found the thread pretty interesting as it combines global politics, trade, military interventions, supply chain logistics and hard computer problems in one.

I’m surprised there is no liability for facilitating scams / fraud on the ad networks given that they’re making money on it.

It’s likely that the grade school math that was referenced is GSM8K https://github.com/openai/grade-school-math which LLMs perform poorly on.

Yeah! Back in the day, the Rhino implementation of JavaScript had serialisable continuations on top of the JDK. I wrote a workflow engine with them.

There are some interesting challenges:

- deserialising them into the same environment

- hooking in dependencies

- for time independence, how do you fix a bug on a process that has been running for 2 weeks and has got 2 more to go

Azure stateful functions solved this by instead of serialising the stack, they just start the process from scratch and then replay every external request / response back into them until they get the process back to its last instruction.

This is pretty neat.

Somewhat related: I’ve been intrigued by serialisable continuations forever. The idea of machine and time independent control flow in a regular programming language could be quite useful eg:

    response = send_email_and_wait_for_response(“hi!”, timeout=“4d”)

Why don’t they run detection on the output and block it if it violates the rules with some degree of certainty e.g. in this case it would be an exact match?

The value to humanity of autonomous organisations optimising for profit is not super clear to me.

I suspect we’re going to see a lot more of this kind of thing e.g. this is currently at the top of GitHub trending

https://github.com/Torantulino/Auto-GPT

I’m not sure it ends well.

Yeah. Pretty high stakes. If we assume that work is already underway to create unaligned AIs, what’s the alternative?

Cat may be out of the bag. I was wondering today whether the race should be on to be creating an benevolent AI to defend humanity from the less benevolent ones.

Looks like chatgpt was trained on stackoverflow

https://ai.stackexchange.com/questions/38660/was-chatgpt-tra...

I wonder what its contribution was to chatgpt’s ability to answer coding questions. And what happens if chatgpt (or similar) displaces it.

Also interesting is that SO has banned chatgpt from answering but I suspect that will be another shadow “AI vs humans” war.

It will be interesting to see how it handles transitive dependencies through a supply chain you don’t control. Same issue as “this 3rd party npm package depends on an insecure version of X” but at a function level.

Thanks! Updated a bit

- text and image generating AI is hugely disruptive to education and content generation

- Inflation eases, stock market recovers

- A Twitter alternative breaks through.

- One of the big tech companies joins the fediverse (Microsoft most likely)

- TikTok starts to falter - additional regulation, banning, lack of profitability and decline in popularity take a toll

- A fully autonomous drone is used in combat (edit: sounds like it may have already happened so adding “and kills a civilian”)

- Chinas ascendancy continues to slow, Xi’s popularity wanes

- We discover evidence of extra-terrestrial life

- someone “marries” their virtual partner (edit: already happened… so perhaps just a lot more of this)

- From last year: First AI generated novel makes best seller list. Arguments about how ‘directed’ it was

- From last year: Interactive Porn AI startups start doing very well

- Crypto is more regulated

I read the article and if I think it’s interesting or the comments might be interesting, I upvote.

Doing technical sales helped me the most as the incentives and feedback for effective human interaction were very clear.

Doing customer support was great for acclimatising to talking to people who were upset.

Improv has been excellent for commoditising awkwardness in a safe space and is just a lot of fun.

Mainly it’s the same as getting good at most things:

- do it often

- get good feedback

- have good mentors

- learn from those who do it effectively

The economist has a good breakdown https://www.economist.com/graphic-detail/coronavirus-excess-...

That’s amazing. It does rely on a level of comfort with a fuzzy error budget.

I wonder if measuring engagement and optimising for engagement should be separated. I’m glad that there is lots of engagement on HN but I’m also glad that HN chooses not to implement a number of engagement optimisation features as well as downranking topics that result in high engagement flame wars.

Oh no. I visit the trending tab almost daily. Any suggestions for alternatives?