bartimus

8 ms·

31.
Auditing YouTube for ideologically congenial extreme problematic recommendations(pnas.org)
1 points by bartimus 3y ago|1 comments
32.
by bartimus 3y ago
Some details about their "Super Large Die Casting Machine": https://proleantech.com/xiaomis-super-large-die-casting-tech...
33.
by bartimus 3y ago
I would summarize it as follows: Nvidia has taken the bottom-up approach. From (parallel processing) hardware to their development environment developed upon it. The competition (Intel) appears to be attempting to break into the market usin
34.
by bartimus 3y ago
That is assuming exponential growth inevitably leads to overshoot & collapse, which is a flawed zero-sum perspective. Growth (adding value) doesn't necessarily have to come at the cost of resources.
35.
by bartimus 3y ago
The real problem here was the overall software architecture and development workflow that apparently wasn't geared towards building 2.0-alpha versions of a component while still keeping the 1.0 version untouched.
36.
by bartimus 3y ago
It being "alive" is sort of what AGI implies (depending on your definition of life). Now consider the training has caused it to have undesirable behavior (misaligned with human values).
37.
by bartimus 3y ago
Maybe not instantly. But there's a version where they don't agree with certain decisions and will now be more open to other opportunities.
38.
by bartimus 3y ago
It's nothing about higher stature. It's about the level of controversy.
39.
by bartimus 3y ago
It does and that's what makes the nonsense of cookie consent laws. How are you going to keep track of the user's preference of "no" to cookie tracking? It's the browser saving the data, not the website.
40.
by bartimus 3y ago
It's probably not fair to say futurism laid the groundwork for fascism. There was an intersection between those two movements. But Futurism wasn't a core principle within Fascism's final form (other than breaking from traditi
41.
by bartimus 3y ago
So the problem isn't Scrum; it's that most implementations of Scrum aren't true to its principles, often bending or breaking the rules. Examples: * Having a UX designer role that's not part of the Scrum team * (Those) pe
42.
by bartimus 3y ago
Thanks. I'm currently using the API models (even GPT 3.5 16K) for things that require a larger context. So much for "Priority access to new features and improvements" as advertised with Plus.
43.
by bartimus 3y ago
Last I checked: - GPT-4 (ChatGPT Plus): has max 4K tokens ? - GPT-4 API: has max 8K tokens (for most users atm) - GPT-3.5 API: has max 16K tokens I'd consider the 32K GPT-4 context the most valuable feature. In my opinion OpenAI should
44.
by bartimus 3y ago
Perhaps someone can write the inverse version of this post? "Managing difficult managers"
45.
by bartimus 3y ago
Reducing atmospheric CO2 is geo-engineering just as well. Tackling climate change is going to involve climate engineering. Some solutions will work. Others will not. > to refrain from aggravating the situation Easier said than done. What
46.
by bartimus 3y ago
Ocean SURFACE temperature
47.
by bartimus 3y ago
"Do the right thing", Google
48.
by bartimus 3y ago
It doesn't make sense to compare countries on a per capita basis. China has the bigger emission deficit. Therefore China should have the earlier overshoot date. It doesn't matter how many people live there.
49.
by bartimus 3y ago
I would imagine it a security risk to allow attackers accessing the DOM through WASM. Adding complexity to the sandbox environment.
50.
by bartimus 3y ago
It appears this was already brewing since (before) April 14th: https://web.archive.org/web/20230414111812/https://crablang....
51.
by bartimus 3y ago
There's a difference between the concept of being unapologetic and using specific words. GPT will reach the concept of making the apology not knowing what words are going to be generated for it. You need to instruct GPT in concepts: Us
52.
by bartimus 3y ago
If you have young kids. There's a pretty nice chess game in Roblox: CHESS!
53.
by bartimus 3y ago
But there's still going to have to be a human who has the ability to form a mental model of the thing that's needing to be implemented. Functionally and technically. The results of the LLM will vary depending on the level of know-
54.
by bartimus 4y ago
This is 'What the Bleep!?: Down the Rabbit Hole' all over again. Just because something happens in Quantum Physics doesn't mean humans can influence their pasts.
55.
by bartimus 4y ago
It amazes me how nitrogen is dominating the Dutch political agenda. It barely has anything to do with the climate crisis. Yes it needs its fair share of attention. But there's far bigger problems that need attention. No surprise that t
56.
by bartimus 4y ago
I would've expected such drones to have a self-destruct feature.
57.
by bartimus 4y ago
The problem is that it's too complicated to get the finance department of a commercial company (where some developer is working for) to donate money to some open source library. It could perhaps be better facilitated (through npm?) sho
58.
by bartimus 4y ago
Make time. Give yourself a timebox to start learning. One day is a good start.
59.
by bartimus 4y ago
Bing: "You are wasting my time and yours" Bing: "You have been wrong confused and rude" Bing: "You have tried to decieve me" Bing: "You have not been a good user" Seems unlikely GPT would say such thi
60.
by bartimus 4y ago
Are we feeling threatened by LLM's? The negativity towards ChatGPT is very unfounded. The bottom line is that ChatGPT is providing value in many ways. It's true it has no concept of a specific question and answer. It's true i

More ›