amf12

19 ms·

181.
by amf12 4y ago
What is an easier solution?
182.
by amf12 4y ago
I don't understand the negative sentiment with this. Do people expect employers to assign everyone a big desk, which is not completely open office and also have the flexibility to come to the office 1-2 days a week? I hope there are ex
183.
by amf12 4y ago
I think the difference being, "the hidden" charges are usually not optional in other cases. Airbnb for example, hotel resort fees, ticket "pickup fees", etc.
184.
by amf12 4y ago
> because the fees are hidden and the total cost appears lower Consumers still understand the total cost, albeit at the end. We aren't tricked into accepting it most of the times. Take Airbnb for example. Consumers compare and find
185.
by amf12 4y ago
Don't worry about PERM right now. You are in a better position because you can get on your wife's dependent visa. Once you find a job, wait until they can get your PERM filed. Once that is done, you can use the unused H1 time peri
186.
by amf12 4y ago
This already exists btw. Have you checked out Google Workspace Individual subscription?
187.
by amf12 4y ago
Sure, but I'd say the benefits outweigh the lack of choice about the infrastructure. I frequently talk to my friend about work, and I realize most of the problems they face - even at good tech companies - are something I never have to
188.
by amf12 4y ago
Google Workspace has this feature. It's mentioned in the post.
189.
by amf12 4y ago
I agree. Its not releasing products that's a problem, its having a long term vision and commitment. I always see two conflicting takes on HN. That Google abandons their products or they haven't done anything recently. Not every pr
190.
by amf12 4y ago
Microsoft was in that position not too long ago. Satya turned it around. Google also needs someone to turn it around.
191.
by amf12 4y ago
> Do you think this is more a reflection on the fact that the more mature company has already made these decisions for you Not really. There are many competing tools internally too, and you have to choose the appropriate one. One thing t
192.
by amf12 4y ago
No, because its "symbolic". Customer outreach is important, but just talking to customers for an hour may or may not help change anything. Maybe have actual outreach programs other than "one hour, and once".
193.
by amf12 4y ago
Those are just fancy words for "improve productivity for businesses and users", which is not that generic.
194.
by amf12 4y ago
> This is completely against the ethos of the internet. The internet was always about new ways to find and organize information I disagree this being applicable to Search. Granted, SEO spam is ranking higher, but a lot of the "inter
195.
by amf12 4y ago
I think this was Google. Google showed that it's working on LLM chatbots in previous I/O. Then, the "sentient thing" about LaMDA blew up in June 2022. It's quite possible OpenAI thought Google would release it soon
196.
by amf12 4y ago
> They should have been experimenting with this stuff in public for years now, but they've kept it under cover. And yet, if they did that, and it didn't go anywhere, it would have been another addition to "KilledByGoogle&q
197.
by amf12 4y ago
> Google looks like the IBM to their Microsoft right now. I really dislike how Microsoft is being considered to "be better at AI" recently. No, they just integrated another AI solution in Bing. They haven't released any bi
198.
by amf12 4y ago
I don't think it's the problem with lack of innovation. Google has innovated in a lot of areas - Transformers (the foundation for ChatGPT), Stadia, Project Starline, and some the public may not have heard of yet. The problem lies
199.
by amf12 4y ago
Do you remember how a Google employee thought LaMDA was sentient and tried to hire a lawyer for the LLM? It's the same thing here. It's just generating words.
200.
by amf12 4y ago
Correlation is not causation. It's also likely that there was more awareness in urban areas, access to mental healthcare, schools that could look for ADHD synonyms. Moreover, ADHD was also a thing 30 years ago when there weren't a
201.
by amf12 4y ago
You are also assuming these are purely linked to environment/society. Some DSM symptoms are likely based on environment and coping mechanisms developed. I do think that current lifestyle makes ADHD worse, but there hasn't been any
202.
by amf12 4y ago
That's a reach. Yes, we don't have biological testing methods for these conditions yet, but that doesn't mean DSM is like "Salem Witch Trials".
203.
by amf12 4y ago
They are trying to suggest ADHD isn't really a thing and the usual "everyone has a little ADHD in them". So stupid. The lack of awareness that other people can have different experiences is just annoying.
204.
by amf12 4y ago
It could be more frequently diagnosed, but also under-diagnosed at the same time. More diagnosis are due to more awareness, but there isn't data to prove it's been overly diagnosed in people who don't have it.
205.
by amf12 4y ago
ADHD isn't just "stimulated by different things" though. It's lacking the executive function to maintain that stimulation. Someone with ADHD might be stimulated by a thing and then get bored very soon and move on to anot
206.
by amf12 4y ago
I think it depends a lot on what your regular searches are. My recent searches were: a local restaurant, find which streaming provider has a show I heard about and it's ratings, look up website for a business, understand some concept,
207.
by amf12 4y ago
I think this highlights a big limitation of LLMs. It doesn't "understand" the text. It's just piecing words together based on the probability. In this case, the key word was "its", which it ignored - but makes
208.
by amf12 4y ago
Unfortunately, this was a very specific fact that was difficult to catch. Nobody realized for the first day that there was a mistake. It doesn't absolve them though. I'd have expected them to verify everything in the response befo
209.
by amf12 4y ago
I doubt that. Current models have been trained using curated datasets and fine tuned by humans. Going forward, I doubt it would be trained on random web data, but only from good sources. Even if the "source" uses text generated by
210.
by amf12 4y ago
+1, this is a great start but we are a long way to go for it to be "useful". My fear is that trust will go down if problems aren't dealt with - specifically hallucinations. At the current stage, it's super useful for cre

More ›