Just a regular Joe.
We don’t know that… I also don’t think the current LLM architecture will achieve human-level thinking, but I do think that people will use agents & LLMs to try a 100 billion or so times to get us there… we’ll likely see many different architectures in the coming years as a result, and some that are close enough to be bloody scary (and some different enough to be even scarier!) We might very well still have LLMs as part of the bigger system.
They represent the values of the American people
I don’t get the impression that most American people vote based on values, but out of fear of “the other party” (and its voter base). Meme driven democracy.
It’s always amusing to see US folks visiting other countries and realising how things mostly just work. Even in Switzerland, with the world’s second most expensive healthcare spending per person.
Read: Lawmakers continue to fail to properly regulate the healthcare industry, leading to uncertainty and litigation potential, therefore nytimes points the finger at AI. Another day, another dollar, in first world USA.
The government can choose to sue (civil) or prosecute individuals/executives (criminal liability).
Assuming there is some legitimate explanation (I’ll be interested in hearing this one!) they likely won’t, but will now establish higher level reporting channels with OpenAI and Anthropic, and perhaps check their important email boxes more often regardless. ;-)
If Microsoft lets the patent sit idle and refuses to license it to others, it would be a very good thing for a few years. Let’s see what happens, though.
Current demand for AI seems anything but slow… do you think this hardware is being hoarded by large buyers, instead of being put to commercial use ASAP?
Whether it takes 3 months or 5 years to pay for itself is a valid Q though, and there is an element of risk that these companies take on with large hardware purchases in a rapidly evolving industry. Ultimately profits will need to cover any debts they take on, and there’s only so much private investment money to go around… I expect to see more consolidation as this plays out.
Owning datacenters with secure energy sources seems like a safe bet though, and is probably the bigger initial investment.
infrastructure, which will be radically cheaper when the bubble pops
I wouldn’t count on that unless the popping sound is nuclear. Demand for AI is not going anywhere, even when companies fail and investors lose out.
What will improve prices for everyone is better algorithms and more efficient chips for them, helping supply to catch up with demand.
Too many lawyers already do half-assed work, and they have already been doing it faster with LLMs. Laymen usually won’t know the difference.
If specialist skills and tools can lead to less shoddy results in spite of endemic human incompetence or similar, it will be better for the average consumer of legal services.
And in the hands of competent and engaged lawyers, better tools are a huge boon.
A game changer (already) is that laymen can often now avoid lawyers altogether as complex laws, contractual terms and procedures can now often be explained in plain english/dutch/swahili by chatbots, helping them to make better and timely decisions.
Note: This is a world-wide phenomenon, with judges, paralegals, prosecutors, police, and end users also reaching for or being impacted by AI these days.
The claude shrine is next to the shrine to the flying spaghetti monster in the games room. The two sides compete regularly in billiards and Foosball (Fußball / table soccer).
News at 11!
Once again, civil (damages, and likely already settled out of court), and not not criminal (a much higher bar, with prosecutors having to prove criminal intent or negligence, which would have been a flop IMHO).
On the sale itself - HF’s owners believed they could make a killing by selling at the right time for the right price. This was it. They had no trouble saying no to lesser offers in the past, and they said yes to the best one. Personally, I’m happy they didn’t go with Microsoft.
The hat suits you… it would look better with a feather though.
I present you with your official tin foil hat. May you wear it proudly!
“My agent hacked hugging face” is almost certainly not a civil legal defence in the US, and I wonder why so many people seem to suggest that it is/was (willful ignorance, is my suspicion).
My bet is that OpenAI offered to compensate HF in one way or another (or already does) and an agreement was made.
OTOH a criminal case would likely be an uphill battle, unless criminal negligence could be established.
There are a lot of people with limited agency over their own work or personal lives, and are at the receiving end of other people’s decisions.
Resentment can build when they perceive outside forces impacting their lives and personal well-being negatively. Sometimes this comes to define them as a person, in a self-reinforcing spiral of doom.
For most well adjusted people in decent work and social circles, we can take a more nuanced view, being both critical and optimistic as needed, and finding satisfaction in making the world work for us.
Consider yourself privileged.
Troops denied their right to vote. News at 11!
why
why get up in the morning, only to have to do it again tomorrow?
why paint a still life, when you can just stare at the real thing for a while?
why not hunt and forage your own food?
why not make your own clothes?
why write boring code, when you can get someone or something else to write it for you?
why contribute features that you think are interesting to a product that you use and enjoy?
why post on social media, rather than doing any of the above?
It is dumb to just let a non-deterministic prediction machine loose with shell access and important credentials.
You either: verify changes, and verify commands against restrictive whitelists, falling back to manual approvals (that you carefully check) before letting them run, or…
You only enable YOLO in a sandbox with temporary/traceable limited access credentials (still some risks, but muuuuch lower) AND audit resulting code or…
You only give a limited set of tools that can do no real harm and can be audited after the fact… but this is more typical of assistants and predefined workflows than coding agents.
The catch: Good sandboxing requires a security mindset and technical knowledge to not screw it up… it’s just too easy to give agents more access than they actually need, and even the experts screw up too often. It’s the wild west out there…