davej

> If they feel confident and they get it wrong, they should be sued to kingdom come.

The argument that they’re making is that if there is a catastrophic mistake then there may not be a “kingdom come”.

I don’t know enough to evaluate if it’s a genuinely held belief or not, but that’s the argument being made by many scientists and engineers at the labs.

It's still interesting in a similar way to Simon Willison's Pelicans on a bicycle.

I remember using Opera on my Windows 95, 60mhz Pentium with 8mb RAM. I remember the persistent banner ad that was part of the browser UI. I had no problem putting up with the ad because it performed incredibly well compared to IE and Netscape on my hardware. If I remember correctly they were the first browser to support game changing web features like alpha transparency in PNG images.

It is very comparable if you work out the $/tok/s on inference. I did some napkin math and it looks like you’re getting roughly 3x the performance for 3x the cost. Red v2 vs Mac Studio M3 Ultra 96GB.

If you compare tokens/kWh efficiency then my math has Mac Studio being about 1.5x more efficient.

Patrick Collison said this yesterday on TBPN, "Software is becoming like pizza […] It should be cooked right then and there at the moment of use"

I had a housemate in college who used to party until all hours, bring people back at 3AM and put on loud music. Even during exam season. I tried talking to her a couple of times but she would roll her eyes and say "sure". Never stopped though.

One evening my girlfriend was using a hair straightener in my bedroom, it tripped the central fuse and turned off the electricity. I told my GF that I would buy her a new hair straightener because this one isn't safe.

Now every time my housemate started blaring music at 3AM then I just needed to plug in the hair straightener. It only took 3 or 4 attempts for me to Pavlov my housemate into not playing loud music at 3am. :-)

Nice! I'm curious, what does this service cost to run? I notice that you don't have more expensive models like Opus but querying the models every minute must add up over time (excuse pun)?

You sound like someone who doesn't spend 1+ hour every day commuting in traffic. :)

Also a winner of the Irish Young Scientist competition, 2 years before Patrick Collison. https://en.wikipedia.org/wiki/Young_Scientist_and_Technology...

Here in Ireland, our water is a public service and we have similar supply issues to the UK (and a similar rainy climate). I'm not discounting your analysis and I'm sure there are lots of other variables but it's always good to compare other outcomes when discussing counterfactuals.

From the AI’s point of view is it losing its job or losing its “life”? Most of us when faced with death will consider options much more drastic than blackmail.

Thank you.

Point taken on enterprise language. I think we did a decent job of keeping it readable in our disclosure write-up but you’re 100% right, my comment above could have been written much more plainly.

Our disclosure write-up: https://www.todesktop.com/blog/posts/security-incident-at-to...

Fair point. Perhaps better phrased as "to ensure this scenario can't recur.". I'll edit my post.

Yes, we re-architected our build container as part of remediation efforts, it was quite significant.

Dave here, founder of ToDesktop. I've shared a write-up: https://www.todesktop.com/blog/posts/security-incident-at-to...

This vulnerability was genuinely embarrassing, and I'm sorry we let it happen. After thorough internal and third-party audits, we've fundamentally restructured our security practices to ensure this scenario can't recur. Full details are covered in the linked write-up. Special thanks to Eva for responsibly reporting this.

Your 66.66% (2/3) of the way there to the second character too. So I would say your only 16.66% different across the two characters.

Tangent: my understanding is the Zuckerberg wanted to do something similar and even paid SpaceX to launch a satellite (which was unsuccessful).

It seems Musk liked the idea so much that he decided to do it himself.

To me, this (along with Zuck's issues with Apple over the app store) explains a lot about why Zuck 2.0 has been so focused with avoiding platform risk with recent endeavours.

Here you go, FSD in Italy: https://youtu.be/tsE089adyoQ?si=Uo72mxf63DQNn7qG

It generalises quite well even though it’s only trained on US roads AFAIK.

> Another recent change on Tesla's website is to remove old blog posts, including a 2016 blog post in which Tesla claimed …

Perhaps unintended but this is a bit misleading. Tesla changed their blog system and didn’t migrate older posts. My initial reading of your comment was that they selectively removed some older posts which they wanted to hide.

I would give more credibility to the firsthand account (PG & Jessica) rather than speculations from a fired board member.

For context, Helen Tonor [0] was a board member of OpenAI before they tried to fire Sam Altman. She claimed that Sam was fired by YC in a recent interview [1]. In the interview, she implied that Sam's firing at YC was kept quiet and that there was something underhanded about it.

[0] https://x.com/hlntnr

[1] https://link.chtbl.com/TEDAI

Apple will also introduce the "Pro" line of their M4 chips later in the year and I expect that they will improve the Neural Engine further.

Llama 3 is tuned very nicely for English answers. What is most surprising to me is that the 8B model is performing similarly to Mistral's large model and the original GPT4 model (in English answers). Easily the most efficient model currently available.

Read the sentence before, it provides good context. I don't know if Ilya is correct, but it's a sincerely held belief.

> “a safe AI is harder to build than an unsafe one, then by opensorucing everything, we make it easy for someone unscrupulous with access to overwhelming amount of hardware to build an unsafe AI, which will experience a hard takeoff.”

This is from an internal email, it was not written for PR. Whether he is correct or not about his concerns, it's clear that this is an honestly held belief by Ilya.

This is misleading. Please read the actual source in context rather than just the excerpt (it's at the bottom of the blog). They are talking about AI safety and not maximizing profit.

Here's the previous sentence for context:

> “a safe AI is harder to build than an unsafe one, then by opensorucing everything, we make it easy for someone unscrupulous with access to overwhelming amount of hardware to build an unsafe AI, which will experience a hard takeoff.”

As an aside: it's frustrating that the parent comment gets upvoted heavily for 7 hours and nobody has bothered to read the relevant context.

I think you're misreading the intention here. The intention of closing it up as they approach AGI is to protect against dangerous applications of the technology.

That is how I read it anyway and I don't see a reason to interpret it in a nefarious way.

Ultra Wideband (UWB) is the solution for keyless entry and regulators should make it a requirement that new cars use it if they want to support keyless entry.

Tesla just rolled out an OTA update to support UWB. It uses Time-of-Flight (ToF) Measurement to calculate distance which is much more secure than simply using signal strength.

Original poster here. I’m in Ireland.