Keep in mind for Anthropic the goal for them isn't necessarily to solve a problem but to demonstrate/measure the innate capabilities of their model. Many results would be a lot less interesting/impressive for an LLM to accomplish if researchers smuggled in key premises or ideas within their prompt.
So many ideas involving AI just seems to be built off of sci-fi (not in a good way), including this one. Like sci-fi, there are little practical considerations made.
Sci-fi isn't even really about the tech. It's about what happens to us, humans, when the tech changes in dramatic ways. Sci-fi authors dream up types of technology that create new social orders, factions, rifts, types of interpersonal relationships, types of fascism, where the unforseen consequences of human ingenuity hoist us upon our collective petard.
But these baffoons only see the blinky shiney and completely miss the point of the stories. They have a child's view of SF the way that men in their teens and 20d thought they were supposed to be like Tyler Durden.
This is a good point and is why I prefer to refer to the genre as Speculative Fiction - not only is it broader but it better gets at the idea behind this type of fiction. Not just space lasers.
What? Fundamentally, information can only be so dense. Current models may be inefficient w.r.t. information density, however, there is a lower bound of compute required. As a pathological example, we shouldn't expect a megabyte worth of parameters to be able to encode the entirety of Wikipedia.
I thought recycle was the breaking down of the device into its constituents (mostly to recover precious metals or other base materials). In contrast, reuse is where the device is kept mostly intact and used for other purposes such as this. Just breaking things down requires considerable amounts of energy through the sheer logistics of it relative to just reusing it.
There is quite a bit of room between reusing something for its intended purpose and recycling something at the material level. TFA is at neither extreme - the phone is kept intact but it's employed for a much more limited purpose that only really uses some of the phone's components.
It might make more sense to think of in terms of expected value. Whilst the probability may be low, the payoff is probably many times the $250M if their startup becomes successful.
That's measuring desirability--or at least immediate desirability, which is sometimes related but distinct from wealth.
It also quickly becomes meaningless as people risking life and limb tend to flee to the nearest stable country. For example, Syrian refugees for the most part tended to stay close to their home country with the majority fleeing to nearby Turkey, Lebanon, Jordan, and Iraq--or, moving elsewhere within Syria.
Measuring wealth of a country by measuring how many refugees it takes in is like projecting future revenue/success of a convenience store by measuring how many people come into the store during an active-shooter situation.
Going by your metric, it would suggest that a Pole is wealthier than a Brit because Poland has nearly 1 million Ukrainian refugees to the UK's .25 mil.
A similar argument could be made for any safety feature that adds cost to vehicles--literally, all of them. If a death is preventable and adds on a relatively inconsequential amount to the cost of a vehicle, then it is the morally correct choice optimize for safety.
The logic doesn’t scale. You can’t impose arbitrary and subjectives thresholds to gloss over this fact. The obvious conclusion is that safety is one of many moral factors to balance.
We still need people on the ground reporting on these things. AI can't (yet) have connections and conversations with relevant parties. Many interviews would not be possible without a reporter developing a rapport with the interviewee beforehand.
How do you define what is needed? The vast majority of reporting does not affect one's own behaviour at all. For example, consider the Snowden leaks, I guarantee you the vast majority of people did not even attempt to modify their own behaviour at all in light of this new information--despite having read or heard about it.
I'd argue that this reporting was still needed, despite it having very little material impact on one's life.
Almost all news has very little material impact on one's life which is why it is treated as entertainment when targeting consumers. The Snowden leaks was an entertaining story that would be easily monetizable.
Just ask any person who works in teaching or any of the numerous faulty AI detectors (they're all faulty).
Any current technology which can used to accurately detect pre-AI content would necessarily imply that that same technology could be used to train an AI to generate content that could skirt by the AI detector. Sure, there is going to be a lag time, but eventually we will run out of non-AI content.
