- the Cul-de-Sac: AI progress flattens as scaling data and compute, RL and algorithmic improvements hit diminishing returns.

- democratization: LLMs decentralize, mirroring the shift from mainframes to personal computers.

- AI creates new jobs and thus new dependencies for the capitalist class

- Any combination of the above.

- https://news.ycombinator.com/item?id=36563187

- https://news.ycombinator.com/item?id=29936104

- https://news.ycombinator.com/item?id=31504052

"The researchers also explored the possibility that index changes are more predictable today than in the past, leading sophisticated market participants to trade ahead of the events. They concluded that this played only a minor role in their findings, however."

The two main factors being:

"...an increase in migrations over time from the S&P MidCap 400 index, and an overall increase in the market's ability to provide liquidity to those investors seeking to buy and sell around the time of index changes.

The additional liquidity has 'made it easier for everyone to trade, and as a result, the prices move less.' says Dr. Greenwood.

And with more stocks migrating to the S&P 500 from the S&P MidCap 400, midcap funds are selling as S&P 500 funds are buying, leading to 'a wash' in demand, he says.

"

Comments URL: https://news.ycombinator.com/item?id=34147858

Points: 4

# Comments: 2

That makes at least two of us. :) But, I suspect there are quite a few more HN readers who like Kafka's works and literature in general (the first is arguably very correlated to the other), judging from how often literature related articles appear on the HN front page. Often, when that is the case I also can't help but wonder the same about the correlation between being a fiction and HN reader.

[1] https://bryanafern.files.wordpress.com/2016/12/on-writing-ra...

This is incorrect in general case and it entirely depends on the model that is used to produce word vectors and the text corpus the model is trained with.

For instance, fastText model is trained on words, but also their parts (n-grams), so it should produce word vectors that would be close (in cosine-distance) to vectors of their corresponding typos and partials, even if the text corpus that was used to train the model doesn't contain same typos and partially typed words verbatim.

This is an awesome project, congratulations!

Could you share details about the machine translation engine that is used (or where to find out more about it)? Are there any plans to open source the extension code (with the WebAssembly optmizations that are mentioned in the article)?

Thanks.

[1]: https://news.ycombinator.com/item?id=21509144 [2]: https://news.ycombinator.com/item?id=9015092

Comments URL: https://news.ycombinator.com/item?id=23080216

Points: 2

# Comments: 0

"3. The Early Hackers

The beginnings of the hacker culture as we know it today can be conveniently dated to 1961, the year MIT acquired the first PDP-1. The Signals and Power committee of MIT's Tech Model Railroad Club adopted the machine as their favorite tech-toy and invented programming tools, slang, and an entire surrounding culture that is still recognizably with us today. These early years have been examined in the first part of Steven Levy's book Hackers [Levy] .

MIT's computer culture seems to have been the first to adopt the term `hacker'. The Tech Model Railroad Club's hackers became the nucleus of MIT's Artificial Intelligence Laboratory, the world's leading center of AI research into the early 1980s. Their influence was spread far wider after 1969, the first year of the ARPANET. "

Source: http://catb.org/esr/writings/hacker-history/hacker-history-3...

If a primitive embedding unit is a search query instead of a word, then I assume that a query vector model is trained on a limited dictionary of queries. I wonder if that implies that the trained vector model can encode only search queries that are already present in its dictionary? If not, I think it would be interesting to know more about how the closed dictionary problem was solved.

Each patient would have series of LSD sessions in a regulated environment (usually in a room with an interior that feels secure and comfortable, a psychoanalyst guiding the session included). First sessions were used to test an individual high dose threshold. Once the high dose for a patient was established they would stick to it. According to the book, 'bad' or traumatic LSD experiences are mostly related to the 2nd (BPM II: Cosmic Engulfment and No Exit) and 3rd (BPM III: The Death-Rebirth Struggle), reliving of birth- and prenatal traumas that are suppressed. It takes usually several high dose sessions with a psychoanalytic integration work afterwards in order to 'resolve' traumas. Once they are resolved they don't resurface and the person generally can lead happier life.

Regarding 'bad trips' with lasting consequences in the aftermath: the author mentions that _in some cases_ when a trauma was not resolved or properly addressed in the course of one session the patient could suffer from various psychotic episodes after the sessions was over. As far as I could tell, simplistically speaking, this was attributed mainly to the fact that the traumatic content (normally suppressed) resurfaced and was sort of left uncovered and 'half-processed'. It was stressed in several places in the book how important is that the 'set and setting' is right for an LSD session and how it could be dangerous and counterproductive if LSD is consumed in a wrong environment or not followed up with an appropriate 'psychoanalytic' work after the session is finished.

I had a rather milder form of depression few years ago (with symptoms such as low-self esteem, loss of motivation to work, eat, exercise and pretty much all standard symptoms except suicidal thoughts). One thing that helped a lot to break negative thought patters and to start pro-actively coping with my own issues was attending few Ayahuasca sessions. I'd recommend it only if you find really experienced, trustworthy, practitioner such as Dr. Gabor Mate and if you are healthy enough (health heart, you cannot use antidepressants and other, prescribed or not, psychoactive drugs and you should do specific diet few week prior to it). Some claim that it's worth 10 years of psychoanalysis. I don't not know how much truth is in that claim, but in my case it really gave me immediate relief, sense of a fresh start and valuable insights about why I'm the way I'm. Of course, a real work starts after it when it takes time and effort to integrate these insights into your everyday life (but I felt motivated to do so).