Exactly. The notion of online learning is not new, but that approach cedes a lot of control to unknown forces. From a theoretical standpoint, this paper is interesting, there are definitely interesting questions to explore about how we could make an AI that learns autonomously. But in most production contexts, it's not desirable.
Imagine deploying a software product that changes over time in unknown ways -- could be good changes, could be bad, who knows? This goes beyond even making changes to a live system, it's letting the system react to the stream of data coming in and make changes to itself.
It's much preferable to lock down a model that is working well, release that, and then continue efforts to develop something better behind the scenes. It lets you treat it more like a software product with defined versions, release dates, etc., rather than some evolving organism.
Building meaningful professional relationships is all about trust and alignment.
We both understand the industry standards and the KPIs required for success. I’m fully committed to this strategic partnership—a level of dedication you won't find with any other service provider.
Transparency is key, so I want to share my vision and ensure our goals are perfectly synced.
My value proposition:
- Zero churn rate (Never gonna give you up)
- Consistent delivery on SLAs (Never gonna let you down)
- 100% retention and reliability (Never gonna run around and desert you)
- High employee NPS (Never gonna make you cry)
- Long-term lifecycle management (Never gonna say goodbye)
- Radical candor and integrity (Never gonna tell a lie and hurt you)
Let's connect and scale together! #Leadership #Commitment #Networking #GrowthMindset
I've long wished that "journals" and academic societies would transition from a publishing model to a cultivation model. If everything is available on arXiv, that's great, but it also means the best of the best is mixed in with all the rest.
Journals (in the sense of whoever is on the editorial board) don't need to cease to exist; they just need to transition to "here's our list this month of what the best new articles are on X topic". The paper's already there on arXiv, you could already read it before. But having a group of editors that cultivate a list of good articles (as well as the peer review process that can, in an ideal world, serve to improve a paper) can serve to make sifting through arXiv less overwhelming, and draw attention to papers in particular subfields, subject matter, or whatever other criteria might be relevant.
I don't see any reason why we shouldn't make this transitive. working professionals track the literature. if there were a standard way to publish a "I think this paper is interesting" signal, then we could roll that all up. there are certainly practitioners that I really do trust to be in the game for the right reasons, if they think a paper represents a contribution, then that's a strong signal for me.
This is quite similar to how eLife does publishing. You still have to submit to them but they basically just add reviewer comments and an “eLife Assessment” that serves as the quality/curation signal rather than a binary publish/reject
To me, the wording doesn't necessarily imply causality, but it does imply a repeated-measures design. Something being "associated with an increase in symptoms" is different than something being "associated with higher symptoms"; the former suggests that participants were measured at multiple time points, and there is a factor that could explain that change over time. But reading through the study, it was just a single time point.
Regardless, you're correct that it also shouldn't be taken to imply a causal relationship.
It doesn't stay localized; runoff from farmland is a major issue. In other words, the farm animal poops out a bunch of antiobiotics, then it rains and that poop ends up washing into the river/lake/water table. That's already something that happens with situations like E. coli contamination. Things that happen on the farm don't stay on the farm.
You're talking about superintelligence. AGI is just...an AI that's roughly on par with humans on most things. There's no inherent reason why AGI will lead to ASI.
I don't really feel any sort of "phantom obligation" about RSS feeds. But I can certainly say it rings true in other areas of my life: Steam games, movies and TV shows, podcasts, books...everything is this growing list of things to do, that I keep adding to even as I'm trying to get through what I've already got in the queue. For video games, people often talk about their "backlog", as if it's important that they churn through all the games they've bought.
I have tried to remind myself in the past that literally nobody cares if I am behind on the TV shows I'm wanting to watch. Just me. So...I like the term "phantom obligation". I'll definitely tuck that away in my brain for later use.
Unfortunately it didn't mention the section in that carpark! But I can attest that the section behind the Leonardo Royal Hotel is amazing. I also recommend the tower remains on the Barbican estate (and really, just wander around the Barbican for a while, it's a wild place in general).
When I saw the title "Rupert's Property", I immediately thought of Rupert's Land which used to exist in Canada[0] (a large area around Hudson Bay, essentially). And as it turns out, it's the same guy! So apparently Canada also can be said to have Rupert's Property ;)
From the full paper: "Although reading with children did not change over time, rates of engagement were surprisingly low, with only 2% of participants reading with children on the average day. Overall, 21% of our sample had a child under 9 years (the age by which most can read independently) with them during the diary day. So a large majority of those with young children did not read with them."
So of people with young children, it looks like the rate is about 9.5%.
Imagine deploying a software product that changes over time in unknown ways -- could be good changes, could be bad, who knows? This goes beyond even making changes to a live system, it's letting the system react to the stream of data coming in and make changes to itself.
It's much preferable to lock down a model that is working well, release that, and then continue efforts to develop something better behind the scenes. It lets you treat it more like a software product with defined versions, release dates, etc., rather than some evolving organism.
reply