_bypa

8 ms·

91.
by _bypa 8mo ago
> the engineer then gets a new version of the app, pushed to them on Slack on their phone, so that he can then merge it to production. I think it's worse that a CTO doesn't seem to know the difference between a PR and an app.
92.
by _bypa 8mo ago
The collective noun for a group of unicorns in AI is known as a hallucination, as in: a hallucination of unicorns.
93.
by _bypa 8mo ago
With LLMs, industrial sabotage at scale becomes feasible: https://ianreppel.org/llm-powered-industrial-sabotage/ What's truly scary is that agents could manufacture "evidence" to back up their attacks ea
94.
by _bypa 8mo ago
With roughly 400 ms round-trip latency does that even make sense? A car travelling 50 km/h manages to advance by 5 m in that time.
95.
$50 on war in Iran: How geopolitical bets have surged on Polymarket(restofworld.org)
6 points by _bypa 8mo ago|2 comments
96.
Child prodigies rarely become elite performers(economist.com)
143 points by _bypa 8mo ago|151 comments
97.
The Wayback Machine's plug-in to fix the internet's broken links problem(techcrunch.com)
4 points by _bypa 8mo ago|0 comments
98.
by _bypa 8mo ago
obstructionist /əb-strŭk′shə-nĭst, ŏb-/ noun One who systematically blocks or interrupts a process, especially one who attempts to impede passage of legislation by the use of delaying tactics, such as a filibuster. One who deliber
99.
by _bypa 8mo ago
That's a false dichotomy, because transistors and ICs are manufactured to be deterministic and nearly perfect. LLMs can never be guaranteed to be like that. Yes, some things are better when manufactured in highly automated ways (like c
100.
by _bypa 8mo ago
And therein lies the crux: some people love to craft each part themselves, whereas others love to orchestrate but not manufacture each part. With LLMs and engineers often being forced by management to use them, everyone is pushed to become
101.
Human–AI Relationships in Fiction(phys.org)
1 points by _bypa 8mo ago|0 comments
102.
by _bypa 8mo ago
Not at anything resembling the instantaneous global reach of today.
103.
by _bypa 8mo ago
What won't happen in society without social media?
104.
AI alignment is a $200B+ product problem, not a research question(betterhalfai.substack.com)
1 points by _bypa 8mo ago|0 comments
105.
Demystifying Evals for AI Agents(anthropic.com)
1 points by _bypa 8mo ago|0 comments
106.
Why India's plan to make AI companies pay for training data should go global(restofworld.org)
4 points by _bypa 9mo ago|0 comments
107.
Ralph Wiggum as a Degenerate Evolutionary Search(ianreppel.org)
3 points by _bypa 9mo ago|0 comments
108.
by _bypa 9mo ago
The real moat is not being forced to use Jira then.
109.
by _bypa 9mo ago
Thanks! I didn't know the name would ring any bells with anyone here. Colour me surprised!
110.
by _bypa 9mo ago
Not sure about Linux but Deezer is decent. You can even upload your MP3s and listen to them in any device with the same account.
111.
by _bypa 9mo ago
The easiest book that's not popsci but actual physics is Barton Zwiebach's "A First Course in String Theory". It does not presuppose a background in QFT. It does require you to know quantum mechanics. Mind you, it's
112.
by _bypa 9mo ago
https://ianreppel.org
113.
2025 Berggruen Prize Essay Competition Winners(berggruen.org)
2 points by _bypa 9mo ago|0 comments
114.
by _bypa 9mo ago
* LLMs, not AIs. AI has mostly never been about predicting the next words only.
115.
by _bypa 9mo ago
I'm still not sure how that's different from Deep Research mode in most chatty LLMs. The collaboration bit is perhaps novel but I have no personal use for it.
116.
by _bypa 9mo ago
It definitively became so when Meta (then still Facebook) acquired WhatsApp in 2014, even if they were to open source it.
117.
by _bypa 9mo ago
> Huang said the rhetoric is "scaring people" from making investments in the improvement of AI. And yet investors and companies are pumping $61 trillion into data centres fuelled by AI, though mostly LLMs. It seems people have
118.
by _bypa 9mo ago
What exactly makes it suitable for deep work and research workflows?
119.
by _bypa 9mo ago
> Is summarizing by a human much different? One thing I have noticed and drives me up the wall with AI-generated summaries is that they don't provide decent summaries most of the time. They are summaries of an actual summary. For in
120.
by _bypa 9mo ago
Isn't it more telling that a Google engineer uses the competitor's LLM rather than their own in-house one?

More ›