May be we should have inverse of all major actively managed funds. Per Buffet theory, this should always outperform market while not having limitations of passive ETFs and much harder to manipulate.
It is not as old as time. It is a very new phenomenon where both sexes now work towards equal earning power, seek absolute freedom without dependencies while child care is professionally commoditized. At the same time, entertainment is not only plentiful and cheap but typically of higher quality than any marriage can provide beyond honeymoon period. Add on the top of this very extreme cost of divorces (in state of CA it translates to lifetime slavery) and most young people very acutely aware about this as half of themselves comes from divorced families. They innately understand this "love" thing is transient, expensive and very time consuming.
From eq 2.2, additive part is usually in few 10s of millions. So, for N > 1B, approximation should be good but it doesn't work. For example, GPT3 inference flops is actually 3.4E+18 so the ratio is 19,000 not 2.
For this work, there should be equal rewards for ICs and managers. That is, ICs should get equal probability to get to level X in the company as managers. This actually used to be the case in old big tech, in early days. ICs can get to all the way to Director and Distinguished level with equal probability of their managers. But during last couple of decades everything changed. Big tech career ladder now states that merely making impact is not good enough but one needs to "scale" the impact. This is sheepish way of saying that one must be manager in order to claim "scaling". Consequently, there no higher level promos in big tech if you are mere IC. Unless this is solved, everything else is just talk and wishful thinking.
This kind of no information exec-speak is very painful to read. I had to pass it through ChatGPT to summarize. THe whole article can be summarized in 1 sentence without any loss of information: "Google will name it's chatbot Bard and it be based on Lambda model, initially small size". No link, no other details, basically a lot of words to say nothing at all.
Definitely not a good name. The first similar word that came into mind was "lard" and then "retard". Also, looks like they will release smaller model to save on compute. This doesn't look like a good strategy. They need to come out solid and strong over competition, not watered down.
Search has particularly gone south last 3-4 months. Quotes and exact phrasing is now a necessity and I find myself rather use ChatGPT.
What bothers me is complete stagnation. I can tell you 5 things that must be fixed in Maps in 5 mins but hasn’t been for a long time. People who are working on it full time should be doing better.
Just had ChatGPT write code for scipy and saved at least hour (it ran correctly as-is). $42 is way too cheap for people in software development at least where hourly rates far more than compensates such costs.
Huge problem is that restaurant owners are usually extremely bad at economics. They think that defaulting to 22% tip (in addition to increasing prices by 50%) will sustain them. In reality many customers avoid sit down restaurants purely because of this.
A decade ago, I had calculated that a startup would need at least $1B (but more likely $4B) of capex to match Google if trying to use conventional crawl/index tech. With conventional tech, quality upper bound is more or less fixed so you are forced to match breadth to compete which is super expensive. With model-based index, you need less than $100M capex to turn on the proverbial flywheel because you can now sacrifice breadth for very different notion of quality. This is a game changer because we will soon see dozen or so strong startups which will chip away the search share, each trying to leverage, cost optimize and improve this tech.
That is exactly the problem and indicates a lack of grand vision. Releasing hodgepodge of internal projects is not same as rebuilding AI-first products.
Moving out of social will turn out to be their big mistake. They should have stayed persistent. The data is the new king and absolute differentiator in the world of AI. Facebook is fortunately sleeping at wheel. Musk is quickly realizing how important this is and have already turned off 3rd party access.
This is old myth. 40% their revenue now comes from other properties such as YouTube, gmail, GCP, Android, maps etc. They are fairly comparable in diversification for rest of the big tech but still little way to go.
I highly doubt if this specific court will make decisions based on what’s good for users or society. They are dead set on philosophy that if it is really that important for users or society than lawmakers should be enacting laws as opposed to Supreme Court sneaking in the protections. So, these people are pursuing wrong line of argument and likely fail. They should instead be pursuing rigorous argument that original intent of Section 230 was to provide the protection they are asking for.
People forget about those days. We used to get daily news about % of ICU capacity left. Videos of people on respirators. We didn’t knew what this virus was. How bad things can go? Streets in China getting piled up with bodies in bags. People complaining that US govt is not as effective as Chinese. People worried about covid deniers, mask opposition. Entire Twitter population updating their profiles with masks. Fear of long covid. News of unknown impact on kids. No one knew how things would turn out. Most believed that that wasn’t time for debate because there was simply no data. People can argue either way and there will be always groups latching on other side. We needed unity and caution until we figured things out. That’s how efforts against covid misinformation started. In retrospect, it looks like suppression of freedom of speech. It wasn’t until vaccines came out and multiple dosages freed us. Now we look back and worry about that suppression of speech issue and packaging it up politically. I am not on either side but I think based on data we had, lack of vaccines and possibly safety issue for entire populations, people had to do what they had to. I don’t think their actions were politically motivated or ill-intended.
This is very First world view sans min wage population. In developing countries, it is always a pricing problem and people don’t care about convenience. When you have only just barely enough money for food and rent, you are not thinking about great service vs zero price trade off.
Zuck immediately needs to hope on the AGI train before it leaves the station. He needs to pivote from Metaverse craziness. He needs to come out for making a mistake and course correct ASAP.
He has massive walled garden of data that no one else has. He had MetaAI folks who knows how to build LLMs. He has AppliedML team who knows how to productize AI. He now even has better reputation for saner policies and privacy, thanks to Twitter. He should be able to spin up AI based search engine in 6-12 months. He can produce AI based Office in similar time frame. All the stars are aligned for him if he has courage to course correct.
People who wants to create their bubble chamber shouldn’t be in open social networks, IMO. Sure, you can say that it’s personal preference. Block feature is like a knife and should be used responsibly when someone has malintent and harassment. It should not be used to suppress opposing argument in respectful debates. To add insult to injury, Lex continues to deceive the public that he is all about loving everyone and respects all sides of arguments. At least, this behavior is damaging if a person is influencer and can impact public policies. It definitely isn’t acceptable behavior from future Twitter CEO.
That's what he says but then in his interview he just praises the person and then plays softball of same old questions like meaning of life, is universe simulation, what will AGI do etc. Few minutes in his interview of Kanya West was the only moment I recall he tried to diverge from this modus operandy.