Maybe that's the problem: it doesn't suck enough?

What's with the black leather jacket he always wears? Does he think he's the Fonz? Or a Ramone?

Essentially, his wife and daughter made the choice:

For a bit more on the subject:

Without science, a hypothesis never gets rigorously confirmed and mainstream enough to ever be considered a theory.

DuckDuckGo allows you to do this. On their homepage you can choose between a regular search or an AI search. I have my default page set to no AI.

But with a sufficient number of fiber-seeking backhoes* censorship will always be possible.

*"fiber-seeking backhoes" being a stand-in for whatever disrupts the physical layer of the network stack. Substitute radio jammers, power disruptions, etc. as appropriate for your target's network.

Political appointees should be public, they are accountable. The rest of civil servants are no different to you doing your job at your place of employment.

Civil servants are paid by the tax dollars. The public has a right to know who their tax dollars are going to.

In my state there is a web site where you can look up the salary of those who work for the government. It includes the person's name and their position title.

The technical person would say it's a collection of protocols and the hardware and software that implement them. The information that happens to be hosted there, other than of course the information necessary to keep the internet functioning (routing tables, DNS records, etc.), is not "the internet."

There is no need for the internet as defined above to die, unless something better has displaced it and almost nobody is using it anymore (BITNET anyone?).

According to the memo, metrics such as AI adoption rates, token consumption, code changes, and code reviews become "deeply flawed" when they are disconnected from actual results. Turning them into targets, Amazon says, encourages wasteful behavior that clashes with its emphasis on "frugality and simplicity" and can raise both compute bills and software complexity.

It's almost as if they're just now realizing Goodhart's Law exists.

Ignore network traffic until security self-defenses are in place.

Yeah, you can have a buffer overflow anytime, and they can be bad, but the situation explicitly pointed out an "invaders not recognized as evil before armed guards are posted" situation.

If this had been a bad buffer overflow that occurred AFTER the defenses were in place, maybe it would've resulted in a compromise, or maybe the defenses would've rendered it ineffective.

... the machine can be affected by outside input before it's security/self-defense system is ready?

Oops.

"Of every 2 million sold, less than 1 million are affected!"

In June, over $500,000 worth of bourbon was stolen out from under the nose of a warehouse in Phildelphia. The first paragraph:

Later in the story, the explanation for how this happened is nearly identical to these tech thefts:

Newman said investigators believe the theft may be tied to a "complex ring" targeting high-value consumer products. The perpetrators appeared to impersonate an existing trucking company, using a company name, phone number and other details to gain trust and secure the load, he said.

"They basically try and look like, sound like ⦠a legitimate trucking company," Newman said. "They spoof the company name, they spoof the trucking dispatcher."

In the company's news release, it said the theft appeared to have been carried out with knowledge of "logistics operations and product movement schedules," suggesting planning and coordination.

If - and this is a big if - Anthropic becomes and stays the industry-leader in LLMs AND if - and this is the second big if - the LLM/AI industry as a whole grows way faster than the rest of the economy - then Anthropic's stock appreciation will likely well-outpace the LLM/AI sector and the market as a whole.

That's the "big reward." The odds of that happening with Anthropic or any other particular AI/LLM company, or any AI/LLM company at all for that matter? Not big enough for me to want to play.

It's high-risk, high-reward, much like dot-coms of the 1990s.

There's also the "known unknowns" of regulatory risks both on the "what will AI be allowed to do" front and the regulations on power-hungry AI data centers themselves. We know these are categories of risks, but we don't have a good handle on how likely any given outcome will be 5 years from now. What are the odds that in 2031 America will be like "you want a data center here? Oh sure, no problem" vs. "you want a data center here? Not In My Backyard!" Once the regulatory environments stabilize it will be much easier to price this into "how much am I willing to pay for a given AI company's stock."

Right now, betting on an AI company's success 5 years down the road feels too much like betting on next year's World Series when we don't even know what the MLB Players Union contract will look like, or if there will even be a season next year.