nsagent

Having worked with Girls Who Code and been in many DEI discussions with faculty, industry, and current/prospective students, a lot of the disparities in supposed interest comes down to women feeling like they don't fit into the CS culture — they often feel especially alienated in the introductory courses [1].

Basically, students who were previously exposed to CS education in high school excel in the introductory courses and often downplay the difficulty of the concepts. This is a very male-oriented perspective to take.

That's why programs like Girls Who Code try to address the gap earlier in the pipeline, since it's hard to fundamentally change the attitudes experienced individuals have in early CS courses. Other approaches some schools have tried include separating intro CS courses by prior experience.

Interestingly, studies have shown that women who stick with the CS curriculum perform as well their male counterparts in higher-level CS courses regardless of their initial exposure to CS, though women often think more poorly of their own abilities [2].

[1]: https://newsroom.ucla.edu/stories/cracking-the-code:-why-are... [2]: http://dl.acm.org/citation.cfm?id=3017771

I've lived (not just visited, but spent at least a year) in places as small as Nome, Alaska and as large as Atlanta, SF, Dallas, and Las Vegas. I just moved from rural MA to rural NY (from one small college town to another).

Growing up I yearned with a passion to live in a big city, but having lived a quite a few, I'm much less enamored with city life, at least in the US. Traffic sucks and often where you want to go is spread out geographically, so you're forced to deal with traffic. I also hate that there is so little greenery, except in a handful of places that are super expensive to afford to live near.

That said, cities do offer quite a few positives. Way more things to do and more people to meet.

I think many European cities seem better designed for foot traffic and bikes, with more nature interspersed, than US cities. Having only visited Europe, I feel like I'd much prefer to live in a European style city than many of the major US cities. Then again, since I didn't live in Europe it could be that many of the places I want to go to on a regular basis (like a climbing gym) are in out of the way places. Obviously not all European cities are the ideal (really hated the hustle and bustle in Rome for example), but many more cities I've visited there seem like nice places to live (I even fantasized about moving to Copenhagen for a while).

What's old [1] is new again... without citing prior work. It's not like it's an unknown work. It was published in ICML and has ~250 citations.

[1]: https://arxiv.org/abs/1902.03249

I frequently recommend MIT's missing semester to students [1]. I agree it sucks they don't teach this material, though to be fair they also neglected it when I was in undergrad a quarter century ago. I remember learning it on my own because I wanted to become better. That autodidact proclivity has served me well over the years.

[1]: https://missing.csail.mit.edu/

> Each time I hear (paraphrased): "It is different this time.", I try to imagine the world 25 years ago. Then think: Is this really that different? Most of time, I think, "No, it isn't that different."

Just be careful not to base your conclusion on anecdata. There are legitimate differences amongst generations in addition to the many commonly held misconceptions, e.g. Millennials are poorer than Boomers were at the same age.

I'd you're interested, check out Jean Twenge's book "Generations" for a look at the differences backed by the best available data.

Considering how often my three-year-old niece wants to play around with my Apple Watch, including at bedtime, it seems like it can be just as big of a distraction.

Search online for beware of dog lawsuits in the US as an example of liability. The problem is that these laws vary by jurisdiction, so for example in California those signs do not protect the homeowner if someone is bitten on their land [1]. Michigan State University has a comprehensive overview of dog bite laws in the US indicating where "beware of dog" signs legally protect the owner from lawsuits [2] when someone trespasses.

[1]: https://www.enjuris.com/california/premises-liability/beware...

[2]: https://www.animallaw.info/topic/table-dog-bite-strict-liabi...

The consumer costs are significantly lower for software. For example, the cost of ChatGPT is minuscule in comparison, while the R&D costs are still quite high.

The real difference between the two sectors is that for medicine, you have a captive audience. People are willing to give up their life savings for certain medical interventions. That same cannot be said for software.

I'm not sure what the right solution is, but making 160x the manufacturing cost seems a bit obscene. Not sure what the total cost of R&D was, but at that rate of return those R&D costs have to be recovered relatively quickly and then it becomes a pure profit machine.

A number of people in my lab do research into long context evaluation of LLMs for works of fiction. The likelihood is very high that Moby Dick is in the training data. Instead the people in my lab have explored recently published books to avoid these issues.

See BooookScore (https://openreview.net/forum?id=7Ttk3RzDeu) which was just presented at ICLR last week and FABLES (https://arxiv.org/abs/2404.01261) a recent preprint.

Thanks for sharing your experience at Apple. It's really valuable to understand exactly how decisions are made at big tech companies, especially for sensitive issues like this.

That said, I'm saddened that the approach seems shortsighted, since it doesn't sound like it considered a holistic picture, but rather was based on a specific value judgement. This seems to put other issues, like the whiplash around the approach to CSAM into perspective.

The extra parenthetical is very much clickbait. The doctor received surgery and care from other doctors using a protocol developed in an institute he helps helm. That's quite different than saying he treated himself.

Reminds me of the Louis CK bit about internet:

https://youtube.com/watch?v=me4BZBsHwZs

For those who prefer videos, my PhD advisor puts all his class lectures online and updates his course materials to cover the latest trends in NLP.

Here's the playlist for the current Spring 2024 semester of the course:

https://www.youtube.com/playlist?list=PLWnsVgP6CzafDszSy-njj...

I've also done research in the use of LLMs for RPGs, using Disco Elysium as a testbed (rather than the small made up scenario seen in this paper).

https://aclanthology.org/2023.findings-emnlp.151/

https://pl.aiwright.dev/

Everyone seems to love equation 1. It really sells the work. Great job ;-P

Great job! A lot of folks from my CS PhD program are really into climbing (including me), but none of us has conducted research that explored climbing. Rather climbing is our respite from research.

Anyway, I think this could work really well as an app. Good luck.

You might like, the Infinigram paper then. It was discussed recently:

https://news.ycombinator.com/item?id=40266791

DEI is applied broadly, for example here's a list of demographics targeted for DEI from one of Biden's executive orders [1]:

  The initiative will advance opportunity for communities that have historically faced employment discrimination and professional barriers, including: people of color; women; first-generation professionals and immigrants; individuals with disabilities; LGBTQ+ individuals; Americans who live in rural areas; older Americans who face age discrimination when seeking employment; parents and caregivers who face employment barriers; people of faith who require religious accommodations at work; individuals who were formerly incarcerated; and veterans and military spouses.

[1]: https://www.whitehouse.gov/briefing-room/statements-releases...

In this case, "outreach" means [1]:

  the extending of services or assistance beyond current or usual limits

  | an outreach program

[1]: https://www.merriam-webster.com/dictionary/outreach#dictiona...

It's not about treating students differently. Rather it's about where you spend your limited resources for outreach.

For example, during my PhD I did outreach in both elementary and middle schools where teachers said there were skills gaps they needed help with. The demographics in some of those schools happened to be such that 80-90% of the students were black and brown.

Really great article! I'm definitely going to use that in the future. Thanks for sharing.

Apparently not enough to prevent it from being repeated... One of Trump's former lawyers did it months after the first case made national news.

https://text.npr.org/2023/12/30/1222273745/michael-cohen-ai-...

Yeah, DDG quality seems abysmal for many of my searches now. I'll then switch to Brave, which sometimes finds what I'm looking for. I rarely ever check Google or Bing as a fallback, but when I do it feels like a screenful of ads and it isn't any more helpful (except Google Image search).

Part of the problem seems like a recency bias in search results. I notice sites frequently update pages with new timestamps, but nothing of substance appears to have changed (e.g. a review of something that was released 4 years ago, but the page was supposedly updated last week). So if I do pretty much the same search that succeeded two months ago, but repeat it today, I might not find the useful result I remember coming across.

I'm sure there are a bunch of other issues related to search and SEO that are affecting search quality. It seems insane that the major search providers don't combat this trend by arming users with more tools to tailor their search, but rather steadily degrade the user experience with no recourse.

Honestly, I think if Google was wise, they'd have a skunkworks team rethinking search from scratch (not tied to AI/LLMs) that starts their own index and tries to come up with an alternative to the current Google Search. Maybe they have that already, though I doubt it. I'm sure if they do have such a team, it's intricately tied to existing infrastructure and team hierarchies which effectively nullifies any chance it has at success.

Oh wow, thanks for the heads up! Turns out my recovery key was also invalid... That's something Apple really should have notified people about. These kinds of slip ups without notifying users is terrible.

Responding to something in your visual field unconsciously is very reminiscent of split-brain studies.

https://en.m.wikipedia.org/wiki/Split-brain

> Ex-Microsoft software developer Igor Ostrovsky believes that soon, there won’t be a developer who doesn’t use AI in their workflows.

I'm really curious about this. I honestly don't see a case where I would use a coding assistant and everyone I have spoken to that does, hasn't been a strong coder (they would characterize themselves this way and I certainly agree with their assessment).

I'd love to hear from strong coders — people whom others normally go to when they have a problem with some code (whether debugging or designing something new) who now regularly use AI coding assistants. What do you find useful? In what way does it improve your workflow? Are there tasks where you explicitly avoid them? How often do you find bugs in the generated code (and a corollary, how frequently has a bug slipped in that was only caught later)?

How does this differ from the 2018 NeurIPS paper, Blockwise Parallel Decoding for Deep Autoregressive Models?

https://arxiv.org/abs/1811.03115

the site just needs some subdomains, like https://ai.doomscroller.xyz for an extra bit of charm