dimask

No, they bought a gpt3.5 license, with privacy related terms that are absent to the normal free version.

Imo stuff like this can have applications in a narrower set of problems, where you can also further finetune or simple prompt engineer a model to have a much higher accuracy. I do not think that they can become the bases in which database work is gonna take place in general, but it could be used to automate certain types of tasks that would be too hard/complicated to automate in a classical programmatic way, but not too hard/complicated a real intelligence to be needed to solve.

Not being able to use VPN with certain providers really put me off. What is the reason that I pay so much money if they blacklist the vpn's ips I use because my network is not secure enough? I was not even trying to go through a different country or anything, just the same country I reside and I made my account.

maybe with less intrusive and lighter optical AR glasses that do not cover all your field of view, walking around the city would be more possible. But expecting to go around the streets or even drive with a huge AR headset is unreasonable for our times.

And in principle, it would be possible to have a weather and environment broadcasted through cameras in any place in the world. And there are already VR games that get it quite right.

> I stare out my office window often as I consider code

One of the attractive features of AR/VR for me is to be able to see sth different than the gray sky out of the window in a big part of the year where I live

This is huge. It would have gotten more traction here if the article was titled "putting cameras on babies' heads is all you need". Hope this sets some alternative paths to transformers' overhype.

I interpret "no (you can change the decision later)" as respecting user decision. I interpret "not now" as "you can say no, but we may keep harassing you until you say yes" as very often is actually the case.

Yes but I assume that people do not usually get "addicted" to sending emails, and that feature seems more oriented towards making sense one did not forget sth or clicked send by mistake (has happened to me). What we talk about would be more relevant in social media, both in passively consuming and in participating more actively.

> One, with software, you could delay any action for a set amount of time as such to give your brain time to make a more concrete decision.

Is there a type of software that already does this? Because it is not like it could not be implemented right now, nor that it would not be useful right now for the exact same reasons. Of course it is completely against most corporations' interests, so not sure who would produce that in the first place.

I did not make a value judgement either; just wanted to point that in the end of the day it is all about searching images in stock websites in one way or another.

> we don't see spouses getting PD or AD often

Actually it may seem so [1], though still there is not any conclusive evidence to support a transmission hypothesis really as all this could be due to increased stress and such factors. Also, brain surgeons' increased risk of AD and more reports of associated risks with regard to contamination from brain operations [2] (similar to the article's ones) provide more indications that such a hypothesis is not completely implausible. Though also far from strongly supporting it or anything, as there is no proper experiment design with control groups etc to make better conclusions.

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2945313/ [2] https://www.medicaldesignandoutsourcing.com/report-suggests-...

There are companies like phind that offer copilot-like services using finetuned versions of CodeLlama-34B, which imo are actually good. But I do not know if such a larger model is gonna be used in such a context.

Did you quantise it? At what level and what was your impression compared to other recent smaller models at that quantisation, if so?

> The great thing about gguf is that it will cross to system RAM if there isn't enough VRAM.

Then you can just run it entirely on CPU. There is no point to buy an expensive GPU to run LLMs to be bottlenecked by your CPU in the first place. Which is why I do not get so excited with these huge models, as they gain less traction as not as many people can run them locally, and finetuning is probably more costly too.

The article does not criticise the fact that "something changed", it criticises what specifically changed. The point of dealing with a problem is not to point fingers and find who is guilty, it is to actually find solutions to the problem. And yeah, frankly, having customers complain about a UX is not the end of the world and nobody needs to be scolded about it. They just have to understand what the complaints are about and make a better UX, which they did. It happens all the time.

Generating = using generative models trained on Getty Images and the like.

It seems that the appropriate design skillset was lacking in the R&D group. Also maybe it was a first attempt to make something, and they did not get through a second iteration to improve it.

Why do we expect that skilled SEs are also skilled UX designers? As everything, design requires training. The problem seemed to be such people trained in design were missing from the R&D team, which sounds like management's fault rather than the engineers' in the first place. Then, the management, while correctly identified the lack of design skills, instead of strengthening their R&D team with that missing talent, they put designers in a different group, creating a different set of issues within the company. Seems a case for an overall bad management in my eyes.

How much RAM are you able to set aside for a ryzen igpu?

Though it most probably does not run in on the 1070 but rather on the cpu. It cannot fit on a 1070, it is not about speed, a 1070 cannot run it period.

> Microsoft’s storage reduction is driven by security risks [...]

> Sustainability is an important consideration that drove Microsoft to make that decision.

Why is a university IT's announcement written as if by Microsoft's PR office? The style of the announcement is very weird. Has McGill outsourced all IT support services to Microsoft, thus it is written essentially by them?

Not at all for universities I would say, at least in EU, but maybe here there is just more care in general put into privacy issues and where the data is stored.

But in general it is a trend that IT services are outsourced and that IT departments, and in general administration, in many universities suffer cuts, which is really bad. Apart from the problems this causes, imo this also pauses a security risk as if people's needs are not well covered by services provided, people will find their own solutions without knowledge or consideration to security.

Ummm no? This is a weird take. I want a device to run software, I do not care about their app platforms per se. I have a macbook and barely use the apple store. If they prevent me from running any software apart from through their platform, that is a problem for me.

Moreover, such links can be used to evade email filters that companies usually employ [1]. Combined with the habit of these corporate email services to obfuscate links "for safety", it can make it much easier to get tricked into being phished.

[1] https://www.bleepingcomputer.com/news/security/linkedin-smar...

If one needs the code to be maintained, not supporting may mean that the code may stop being maintained, which may mean that one will stop being able to use it in the future.

This may or may not be related to ethics, depending on what one calls or defines as ethics, but the reality is that free rides always carry a risk, even if there is no ethical obligation to pay for the ride. Of course, that depends on what value one gives to the code, and stuff like the cost of not being able to use it in the future due to lack of maintenance, or the cost of having to put in effort to maintain it oneself. But while there are no such ethical obligations tied to the use of some free software per se, imo there are ethics (or "ethics") tied to the interdepedencies and connections within such a social system in the sense of it being sustainable and able to self-perpetuate.

Well, in the great scheme of things after millions of years nothing matters really, but this is not a reasonable argument for talking about harm done to humans and ecosystems _now_.

YOU SAVED ME

Research software engineering is a hybrid field and fixed roles do not work well. It requires one having a grasp (at some level) of a lot of different stuff, eg software engineerin, the relevant scientific theories, statistics _and_, quite importantly, the culture of scientific practices in a field, in order to make something good. So it gathers a lot of people who, no matter where they started from, often have to sort of converge by learning stuff outside their own discipline.

The problem is not that "physicists write code". Scientists end up learning a lot of stuff and getting good at it (software engineers the same). The problem is that writing software is often left as a job for the occasional phd, postdoc or research assistant, ie people with temporary positions, and in general to people who see building software as a side duty at best, annoyance at worst. This results in no generational knowledge building, being hard to find mentors, less learning and reflecting on practices on how to build software, rediscovering the wheel constantly, too much effort put into . It is not that one graduates as "software engineer" or as a "scientist" and then they know the best practices and everything of their respective fields. People get to learn stuff. Software engineering practices should be in the culture of scientific software building, not simply carrying them from the software engineering world to science, but adapting them and taking it each own idiosyncrasies as a field.

I totally feel this. Since I got my m1 air I have to constantly monitor my laptop's cpu and ram usage, because otherwise I get no feedback at all of what is going on in my computer. I never had the concept to constantly monitor these unless something was wrong. If something was overtaxing the cpu, I would hear the fans. Now, sometimes I only notice that it has gone hot to the touch after quite a while. If I messed up with memory management or forgotten some dozens of tabs open while running memory-heave computations, the OS would start lagging. Now, mostly I can only notice if I check the swap memory used.

> This is a great write-up and I love all the different ways they collected and analyzed data.

> [..] due to inherent biases in the sample set [..]

But that is an analysis methods issue. This serves as a reminder that one cannot depend on AI-assistants when they are not themselves enough knowledgeable on a topic. At least for the time being.

For once, as you point, they conducted a t-test on data that are not independently sampled, as multiple data points were sampled by different people, and there are very valid reasons to believe that different people would have different tasks that may be more or less compute-demanding, which confound the data. This violates one of the very fundamental assumptions of the t-test, which was not pointed out by the code interpreter. In contrast, they could have modeled their data with what is called "linear mixed effects model" where stuff like person (who the laptop belongs to) as well as possibly other stuff like seniority etc could be put into the model as "random effects".

Nevertheless it is all quite interesting data. What I found most interesting is the RAM-related part: caching data can be very powerful, and higher RAM brings more benefits than people usually realise. Any laptop (or at least macbook) with more RAM than it usually needs has most of the time its extra RAM filled by cache.