roca

Google staff aren't allowed to share Imagen results with you, unfortunately.

Keeping GPT-3 behind an API lets OpenAI track how the model is being used and filter outputs they deem potentially harmful.

In the USA, the baby boomer gerontocracy has amassed dominating wealth and power to the detriment of all. It's daunting to imagine a world where they never die. "Do something about it" is much easier said than done.

Another angle: what's science going to be like when no-one ever retires from faculty positions?

It's true that writing code in safe Rust does not make other non-Rust code safe.

Nevertheless, the more code we write in Rust, the safer we will be. Especially because popular kernels and system libraries are often very well tested these days, whereas application code newly written by developers of varying skill levels likely won't be.

> the big problem is that for applications that do need to do non-trivial reference semantics, Rust doesn't really offer much.

For those applications you can almost always encapsulate the unsafe code in a data structure library with a safe interface, such that the library is a tiny fraction of the application code. And in many cases someone else already wrote that library. For graphs, for example, there is petgraph.

> for applications that can easily fit within standard containers and standard allocation paradigms, C++ actually works really well already

Not at all. For example it is still easy to corrupt memory using C++ references. Just the other day I discovered a nasty little bug involving absl::hash_map: someone had written "map[i] = map[j]" which is unsafe if the element at i does not already exist --- it can trigger a rehash, invalidating the reference obtained by map[j].

You seem to be saying that if she had not reigned, that suffering would not have occurred. That seems unlikely.

For better or worse, you very very quickly pile up complexity when programming in a way that novices can't deal with (and experts too, often).

What we're going to find out is whether novice users of image generator AIs frequently get stuck with inadequate results. IMHO the answer is likely to be "no", especially after we've iterated on the tools for a few more years.

A function returning a value that depends on the lifetime of the function's parameter is not crazy at all. Every class getter method that returns a reference to a member of the class does this.

Political representation of different view/value systems is much better in the USA than in China.

> (Though viewing China as a full autocracy with absolute power of a single individual is extremely simplistic)

It's overly simplistic, but less "overly" than it was 10 years ago.

The CCP:

-- Put a million Uighurs in camps

-- Persecutes followers of pretty much all religions --- Muslims, Christians, Buddhists, Falun Gong, etc

-- Flat-out censors anything that offends them in the slightest

-- Puts whoever they want into detention with no kind of due process, from small-time activists to billionaires

-- Harvests the organs of (not previously dead) prisoners

-- Had drones peering into apartment windows during the latest COVID lockdown, broadcasting the literal message "control your soul's desire for freedom"

-- and of course is an autocracy in which Xi Jinping has absolute power

The USA is far from perfect but one thing we learned from Trump in 2016 and Bloomberg in 2020 is that money isn't everything in American politics. The parties do have major differences and bad actors can be held accountable.

I'm not American, but this equivalence that some people like to draw between the governments of China and the USA drives me crazy.

You don't have to be American to see American big-tech companies as the lesser of two evils here.

I strongly agree with this. It's a dangerous tool, and it's easy to imagine the CCP seizing that tool in an emergency.

iPhone Firefox isn't really Firefox, it's still using Apple's Webkit engine (as does Chrome on iPhone).

I'm just saying that declining office attendance is self-reinforcing, not making a judgment whether it's good or bad.

FWIW I think remote work has many good effects. It's let me live in New Zealand and do interesting work, for example. I think it's good for communities. I'm generally very pleased by the explosion in remote working. But after working from home exclusively for six years I'm enjoying being in an office.

There are (at least) two powerful forces that will make "hybrid work" evolve into "WFH" at many offices.

What I saw at Mozilla with a remote-heavy workforce is that declining office attendance is a vicious cycle. The fewer people there are in the office on any given day, the less incentive there is for those people to keep showing up.

You can mitigate that to some extent by requiring people to be in the office on a few specific days (as many "hybrid work" programs are trying to do). The problem then is the company has a lot of expensive real estate that's only being used 3 days a week. That seems wasteful. I predict that as leases come up for renewal, and as they get more used to remote work, a lot of companies are going to look at their office investments and downsize or eliminate them. The super-rich Big Tech companies and the like will continue to spend eye-watering amounts of money on lightly used office space and services, because they can, but I don't see how that's going to work for normal companies.

To a large extent this is what happened at Mozilla pre-pandemic. It was always a remote-heavy group because of its open-source roots, and over time office attendance just eroded away in many locations. I expect this to happen at a wide range of companies as they try to do this "hybrid work" thing.

It wasn't just the original "pedo guy" comment. Musk also hired a private investigator to dig up dirt on his target. He tweeted more, and emailed a Buzzfeed reporter alleging his target was a child rapist. This wasn't just one moment of spite.

You're right. My bad.

So Musk was "too busy" to tell someone "do a background check on this guy" (who was a convicted felon, so not difficult), but not too busy to repeat his claims in emails to Buzzfeed etc.

I'm not perfect, but at least I haven't tried to ruin someone by spreading lies about them to millions of Twitter followers.

Taking your version of events at face value: Musk hired a private investigator to dig up dirt on someone who merely criticized his rescue plan, received some lies from said investigator, and uncritically repeated them to his massive social media following, thus smearing the person concerned ... and this makes Musk a tireless advocate for the truth and social media sanity.

Musk cultists are the worst.

PS, if Musk can so easily be deceived into spreading false information --- even when he, so incredibly rich and smart, hired the source of the information --- why would you trust anything he says?

EDIT: "Musk cultists are the worst" was out of line. I apologise.

Twitter management probably didn't have much choice but to accept the deal. Turning down an offer significantly above the stock price sounds like a trigger for a shareholder lawsuit.

The "cave diver paedophile" incident made it abundantly clear to everyone willing to see that whatever else he may be, Musk is a malicious clown. It was always only a matter of time before that had greater consequences.

It won't be long before paedophiles can produce video avatars of themselves as children. In fact they might already be able to do that.

Explain away to a five-year-old but it's not enough. You will also need to practice safe behaviors and sometimes yell.

Welcome to the metaverse.

Why can't you have a maskless face-to-face meeting? I'm having maskless face-to-face meetings in Auckland all the time.

Yeah. I suspect we're only months, or a few years at most, from automation that's as good or better than these human slaves. Then they can A/B test their way to increasing effectiveness.

An interesting twist will be to pull the voices of your friends off social media videos and impersonate them to you.

Eventually the "industry" will burn through the available pool of gullible "investors" who can be suckered into giving up their money. New suckers are born every minute, of course, but they won't be enough to keep the crypto industry above the background noise of regular frauds and scams.

https://pernos.co does something similar, except that instead of searching for watchpoints/breakpoints we're building a database of everything that happened. That involves replaying different segments of the execution in parallel and also applying different kinds of instrumentation to the same segments, also run in parallel. It's great, we can saturate the biggest boxes AWS has :-).

This is such an obviously valid objection that I'm disturbed the author didn't address it inline.