strin

So meta

My two cents on paid search engines: yes so much awareness around privacy these days. so there might be a potential shift in user behavior, which opens up a big market. however, i am skeptical because people are used to free search engine that surfaces good results.

so privacy is great value prop. but not so much that i am willing to pay for it while living with worse search quality.

Interesting fact. Also all convos lead to bitcoin at some point these days...

lol they didn’t open source the model weights

is the "retrofit" strategy living in the past? living in 2021 it seems a bad choice to buy gasoline cars. Most new cars coming out will have some kind of driving assist (L2 autopilot).

I wonder how clubhouse would monetize its traffic. most social networks had to make the advertisement presented in similar ways to the content, e.g. twitter, instagram, tiktok. but I can't imagine clicking on and listen to a "conversation" that advertises a product...

That works only for static page though. Many modern pages would require you to run a selenium or puppetteer to scrape the content.

Follow a normal SaaS growth curve this will IPO in 25 years!

I wonder if the world would converge to an open standard for social networks: https://news.stanford.edu/news/2014/march/privacy-economy-ap...

This is so accurate as I started my blog in 2018 and the only post I have so far is about the setup https://www.timshi.ai/blog/hello-blog-with-react

Well “starting at 799” nothing wrong with that

> Regardless, leaders and technology evangelists in this field still preach that by 2025, the overall percentage of the pie for voice will end up being about 15%.

What data points is this prediction based on?

> “GPT-3″ is just a bigger GPT-2. In other words, it’s a straightforward generalization of the “just make the transformers bigger” approach

Yes it’s true. But there is a difference between what’s interesting and what works. deep learning (RNNs, transformers, etc.) is usually old ideas applied at large scale with slight modifications. Proving a model works well at large scale (175B parameters) is a great contribution and measures our progress towards AI.

The fact that Bitcoin is marketed as the “new gold” is a double-edged sword. Because of the hype, it’s too volatile to become gold.

> The following year, the lab hired Ritchie, a Harvard grad whose father was a Bell Labs alum, and Ken Thompson, a self-described army brat who, by his own admission, spent at least a month actively dodging the Bell Labs recruiter on the Cal-Berkeley campus

> In our experiments with Transformers, we observed that not all the attention heads utilize their attention span to the fullest. In fact, in a task of character-level language modeling, most of the heads were using only a small portion of their attention span. If we can take advantage of this property during training, we can reduce the computation time and memory footprint significantly

Must read: the structure of scientific revolution https://en.m.wikipedia.org/wiki/The_Structure_of_Scientific_...

tl;dr most of the time science is incremental innovation. and big discovery comes as paradigm shifts, usually first started by a group of “non-mainstream” researchers at the time.

Listen and hear all perspectives:)

> Hinton has said that when he was growing up, his mother gave him two choices: “Be an academic or be a failure.” His family tree is branch-breakingly weighted with scientists. His great-great-grandfather was George Boole, founder of Boolean logic, familiar to anyone who has done a “Boolean search.” One of George Boole’s sons-in-law was Charles Howard Hinton, Geoffrey’s great-grandfather, a mathematician and sci-fi writer who coined the concept of a “tesseract” (a four-dimensional object we can see in the 3-D world as a cube—well known to all readers of the classic children’s novel A Wrinkle in Time), and who ended up in the U.S.

Interesting fact from reading this that normally gets neglected in media.

Sure - you can argue that to improve speech recognition you’ll need to collect ground truth.

But only a tiny fraction of the billion users are tech savvy enough to realize this. Most people are kept in the dark. And we are trusting Apple to be benevolent to not do evil with these recordings.

Aurora Database + auto-scaling #instances = Serverless??

Looks like Elastifile allows users to "mount" cloud buckets to their servers as NFS. This allows them to run old desktop workflows without the need to migrate these workflows to the cloud.

The upside is that Google Cloud is getting serious about the enterprise because most of them still use desktop applications.

The question is how long this will last since the future is the cloud.

Used Google glass for a project three years ago. The biggest question for going to Enterprise is demonstrating the value the solution provides, rather than showing some cool gadgets. I still don't see a killer use case of Google glass.

I can imagine the world relying more and more on unsupervised pre-training approaches, such as BERT and GPT-2. Then we’ll just need a few labeled data to generalize.

Sometimes what users want is what users think of as a solution to their problem in the context they have. For example, users want better taxi services -> turns out ride-sharing is a better solution. Taxi service is a solution to the problem of getting point A to B.

It depends on how "deep learning" is defined. If it's neural networks, the answer is likely no. but a broader definition of feature learning or "end-to-end" learning makes sense. It's likely for AGI what we need is a good, shared representation across multiple tasks, and continual learning paradigm that keeps building up new skills.

Github has been losing the battle in the enterprise world to Gitlab. The latter has more appealing features to enterprise users. So this acquisition in some sense a great opportunity for Github as well to enter enterprise under the wings of Microsoft.

The most terrifying aspect for me is not massive job replacement, but such AI used for robocalls, fraud, ... Imagine you got a phone call that sounds exactly like your mom and asking for help!

This is super cool. I've used mitmproxy 2.0 in my project (http://world-of-bits.com) to create an offline approximation of the website. We were trying to create a digital environment for AI agents to learn to interact with the web. The proxy captures a good amount of HTTP traffic.