Counter point. I would wager people are more productive scrolling 5 minutes through a Facebook post then taking a 30 minute coffee break talking to various coworkers. I would hate this. Also if you’re a developer how would you research something? No stack overflow? No access to forums to solve particular problems? Not sure this is sustainable.
Losing access to language reference docs would be huge. What are they gonna do, save them all locally? Maintain copies of those sites on the company intranet, at the company’s expense? What happens when the next version of Python is released?
This is a real cut the nose the spite the face move. Google would hemorrhage developers.
I mean, Google does index and cache most webpages internally already. So yeah, maybe. But after reading the article it doesn’t sound like they’re doing that.
I mean let’s say they solve that part, sure. Let’s go back to Google’s original intent for this maneuver: they want to beef up “security.”
Ars Technica’s sub-title line says “You can’t get hacked if you aren’t on the Internet.” That is utter nonsense. I’ll take “What is E-Mail?” for 500 Alex. Surely they wouldn’t block EMAIL right? How would they communicate with vendors, partners, governments, etc? Does Google think phishing emails, ransomware, etc don’t work if you don’t have internet access?
Actually, most email malware is staged now, so it wouldn’t work. PDFs with the malware embedded get flagged, so PDFs with a link to the malware replaced them. Even most ransomware is via an external link.
Storing local copies of docs is a thing some companies do. I’ve worked at a couple of places that did that. And when the next version of $foo is released, and the devs get the go-ahead to use it, wget gets executed to make a new copy. Sucks, but that’s the threat model in some places.
LLMs produce text. They don’t answer questions. If the probability of the keywords in the question are being used in correlation with the answer often enough, it might (re)produce the actual answer. But you can never be sure.
Counter point. I would wager people are more productive scrolling 5 minutes through a Facebook post then taking a 30 minute coffee break talking to various coworkers. I would hate this. Also if you’re a developer how would you research something? No stack overflow? No access to forums to solve particular problems? Not sure this is sustainable.
Losing access to language reference docs would be huge. What are they gonna do, save them all locally? Maintain copies of those sites on the company intranet, at the company’s expense? What happens when the next version of Python is released?
This is a real cut the nose the spite the face move. Google would hemorrhage developers.
I mean, Google does index and cache most webpages internally already. So yeah, maybe. But after reading the article it doesn’t sound like they’re doing that.
I mean let’s say they solve that part, sure. Let’s go back to Google’s original intent for this maneuver: they want to beef up “security.”
Ars Technica’s sub-title line says “You can’t get hacked if you aren’t on the Internet.” That is utter nonsense. I’ll take “What is E-Mail?” for 500 Alex. Surely they wouldn’t block EMAIL right? How would they communicate with vendors, partners, governments, etc? Does Google think phishing emails, ransomware, etc don’t work if you don’t have internet access?
Actually, most email malware is staged now, so it wouldn’t work. PDFs with the malware embedded get flagged, so PDFs with a link to the malware replaced them. Even most ransomware is via an external link.
Storing local copies of docs is a thing some companies do. I’ve worked at a couple of places that did that. And when the next version of $foo is released, and the devs get the go-ahead to use it,
wgetgets executed to make a new copy. Sucks, but that’s the threat model in some places.If I had access to a good LLM, that’d be enough for 99% of my research. And the other 1% I could probably do on a phone.
LLMs produce text. They don’t answer questions. If the probability of the keywords in the question are being used in correlation with the answer often enough, it might (re)produce the actual answer. But you can never be sure.
LLMs are not a source for information.