![](https://lemmy.ca/pictrs/image/121e190c-22b0-4198-9a67-51a610ac651f.png)
![](https://lemmy.ml/pictrs/image/q98XK4sKtw.png)
Hard to tell from first glance but my guess would be this is fallout from the ongoing xz
drama. Here: https://www.openwall.com/lists/oss-security/2024/03/29/4
Hard to tell from first glance but my guess would be this is fallout from the ongoing xz
drama. Here: https://www.openwall.com/lists/oss-security/2024/03/29/4
You’re conferring a level of agency where none exists.
It appears to “understand.” It appears to be “knowledgeable. “
But LLMs do neither of those things.
Take this note from an OpenAI dev:
It’s that these models have leveraged so much data they’ve been able to map out relationships between words (or images) in way as to be able to generate what seem like new versions of those things.
I grant you that an LLM has more base level knowledge than any one human, but again this is thanks to terrifyingly large dataset and a design that means it can access this data reasonably reliably.
But it is still a prediction model. It just has more context, better design and (most importantly) data to make predictions at a level never before seen.
If you’ve ever had a chance to play with a model at level where you can control some of its basic parameters it offers a glimpse into just how much of a prediction machine it can be.
My favourite game for a while was to give midjourney a wildly vague prompt but crank the chaos up to 100 (literally the chaos flag at the highest level) to see what kind of wild connections exist but are being filtered out during “normal” use.
The same with the GPT-3.5 API in the “early days” - you could return multiple versions of the response and see the sausage being made to a very small degree.
It doesn’t take away from the sense of magic using these tools. It just helps frame what’s going on under the hood.
We really can vote with our dollars. The issue is that we don’t (I’m point right at myself here).
Don’t buy the things, we probably don’t need em.
Braces for the downvotes.
Teslas are SaaS on wheels. Most EVs are sadly.
In many cases a lot of sites don’t make it clear that they have a conflict of interest.
You want to push a product on me and you’ll get a cut? Cool, but disclose that.
Thanks for engaging!
I should comment on my poorly phrased use of “invite only” -> I mean that from the creators’ perspective in case I fumbled that one. IE, I can’t just decide to share on Nebula, I have to be invited (seemingly by the creator personally? Ref: https://www.theverge.com/23076663/nebula-youtube-creator-business-future-startup-ceo-dave-wiskus).
As a former Nebula subscriber, here’s my hot take: it also has no real community and no chance for exposure to the up-and-comer (IE no way to breakout since it seems invite only?)
I’ve found so many great YouTube channels filled with deep experience and expertise before they “catch on” (and some never “catch on”). The ability to find the small, powerful voice who’s just trying to share knowledge…
I’m not defending YouTube/Alphabet here (as a company they’re no better than any other), I just think Nebula isn’t a great alternative and unless things change, can never be. It’s a walled garden in too many ways (paywall/creator invitations).
In the year I subscribed to Nebula, I mostly watched the same videos on YouTube. If they were technical enough there was valuable discussion attached to the video; on Nebula that’s not the case and not possible. Even if it was possible I can’t imagine people fragmenting their discussion spaces between YouTube and a closed ecosystem like Nebula.
Don’t even get me started with their (Nebula) inability to build a video queue -> wasting time and space on a poorly thought-out implementation of Autoplay was a terrible decision that further pushed me off the platform.
It’s sad, I really wanted to like it. But I voted with my dollars and left.
Under his eye.
While you’re technically right, I don’t see a material difference between paying with cash and paying with data (Verge sign up is free, but it’s still sign up).
Cluster of Pi4 8GBs. Bought pre-pandemic; love the little things.
Nomad, Consul, Gluster, w/ TrueNas-backed NFS for the big files.
They do all sorts of nifty things for us including Nightscout, LanguageTool OSS, monitoring for ubiquiti, Nextdrive, Grafana (which I use for home monitoring - temps/humidity with alerts), Prometheus & Mimir, Postgres, Codeserver.
Basically I use them to schedule dockerized services I want to run or am interested in playing with/learning.
Also I use Rapsberry Pi zero 2 w’s with Shairport-sync (https://github.com/mikebrady/shairport-sync ) as Airplay 2 streaming bridges for audio equipment that isn’t networked or doesn’t support AirPlay 2.
I’m not sure I’d buy a Pi4 today; but they’ve been great so far.
I look forward to the day we can use our “dead” car battery as the battery backup for our home.
64kWh * 0.8 is 51kWh.
Even 40kWh would be a great battery paired to a solar system.
The used car batteries could have great second lives.
Detail transfer is something you get by shooting at a higher resolution and then downscaling.
For example a typical 4k camera will produce a 1080p image that looks significantly more detailed than a 1080p (native) camera (there’s a lot of hand waving here about resolution and lenses but let’s just ignore it all for the sake of the question to on ).
Sort of like how 35mm films transferred to VHS always looked so much sharper with more detail than video shot on VHS-quality equipment.
There’s a lot to unpack here but hopefully it’s enough to kickstart clarifying what they’re talking about.
Sort of?
The big tech companies often pay workers well outside the bands of other firms, particularly when you factor in the equity portion of compensation.
We’re interviewing a Googler right now and they’re going to knowingly take a pay cut to join a company that’s fully remote.
Likely this persons’ move isn’t only about it being remote but also that they’d have more agency in a smaller org.
Salaried.
I think it’s fair to say that I got lucky - you’re not wrong, they needed someone and I happened to have a very specific experience that matched with their need.
I met the team, I knew they liked me, and I really liked them. I also knew it was likely they couldn’t afford the comp I was getting at $lastjob.
My plan going in was to use that (scary to them, but very real) number to negotiate a four day work week at a full time salary. It worked out.
So far so good.
Not downvoted, appreciate you sharing your perspective.
I’ve been successful building trust in remote work settings but it’s a very much about building a narrative that’s much more explicit and communicated in an active way.
But ignoring that bullshit I just typed, I think “building trust” in a professional environment is largely a trap. Not because you can’t trust anyone but that, if you’re building a good team, trust should be implicit. I was hired to do a job, you were hired to do a job, let’s trust that each other to do it.
I think it’s also worth bearing in mind that high trust teams can still build trust, I’m simply advocating for not starting from zero.
Unfortunately so many of the tools and workflows are built explicitly for low trust teams.
As someone who recently started working 4 days, there is absolutely less void in my life. Though the pains of the working days are often amplified because there’s just that much more awareness of weighing each working day against the life I’m building outside.
Negotiated a 4 day work week at my current job.
It’s worth it, but it requires heavy self-moderation to do it at an organization where everyone is 5 days.
AMA.
A bit odd that the article doesn’t mention advertising on cable/sat/fiber/traditional(?) media delivery into the home.
The single biggest draw, to me, isn’t that I can watch when I want (that’s second). It’s not having to spend my time watching ads. Life is just better without someone trying to sell me something for 20 minutes out of every hour.
I’m willing to pay for that privilege.
I value my time - or at least the opportunity to spend it how I want when I’m not making someone already rich, even richer.
Grateful that they don’t. But they have tried to do it with podcasts.
Spotify “pulled an Apple”, bought Gimlet and moved all their podcasts onto Spotify exclusively. I don’t use Spotify and chose to find alternatives. I’m happy I did.
pfBlockerNG at the network edge and ublockorigin on devices.