chrisdbanks

> The brain has always been compared with a highly sophisticated computer.

This is not strictly true. In ancient times, Plato compared the mind to a chariot pulled by two horses, representing emotion and reason. This reflected the importance of horses for transportation at the time.

In the 1600s, Descartes saw the mind as akin to a mechanical clockwork machine, reflective of the clockwork and automatons built then.

In the late 1800s, Freud described the psyche in terms of hydraulics and steam engines - cutting-edge technology.

Each era attempts to explain the complexities of the human mind through the lens of its most advanced creations. But all these analogies fall short.

Is Figma fast? I see huge lag on loading files. Sometimes images don't load for 10+ seconds after the file opens. Based in EU we often don't get thought about in testing.

Yes. Outlier experiments vs incremental experiments. Sometimes you need to break out of a local maxima.

Skill levels follow a power law distribution just like wealth so most people are below average if you think of the average as the mean value. But with skewed distributions the mean is almost meaningless. If you'll pardon the pun.

Workers vs investors shouldn't be seen as a zero sum game. If it is then that's short-term thinking. Simon Sinek explains this all very well in The Infinite Game.

Agree. In fact one problem many have when new to management is actually doing the work themselves rather than coaching others to achieve it.

I'm not defending anything. Just saying that drinkable tap water doesn't exist in many countries, bro. And many people don't even have taps.

No, true, but that's separate to the issue of PFAS in my drinking water making my house worthless. Presumably if there are PFAS is bottled water then everyone's water is contaminated as I'm not aware of anywhere people don't drink bottled water.

You're suggesting that bottled water is worse for me than non-drinking water from the tap? Billions of people in the world don't have access to clean drinking water. You're lucky if you do.

Yes. In plenty of places people use bottled water even to brush their teeth.

I personally use bottled water for cooking. Agriculture mostly doesn't happen where my house is so is unrelated to the house price.

You don't have to drink the water where you live. Where I L live everyone drinks bottled water.

Shouldn't that be "spread like weeds"

Ditto. Recently moved from Pocket to Readwise and it's so much better. Readwise also supports spaced repetition for your highlights. Pocket could have been so much more. It could have surface articles I'd like using a Spotify style algorithm, it could have been so amazing, but Mozilla have killed it. I don't understand how they're so incompetent.

Maybe this is why you've been ghosted so much.

Norway is part of the EEA which means that there is free movement of people and capital. Therefore it is impossible for them to prevent people leaving to other EEA countries. Changing the EEAs 4 freedoms is unlikely.

How much has changed since 2015. With NLP and ML it used to be so hard to create high-quality datasets. It was a case of rubbish in, rubbish out. Not LLMs have solved that problem. It seems that if you put in a huge amount of data into a big enough model then an emergent ability seems to be the ability to discern the wheat from the chaff. Certainly in the NLP space, the days of crowd sourced datasets seem to be over replaced with few shot learning. So much value has been unlocked.

It feels like Google have got to the point where they've actually given up on an algorithm and have many hand curated lists of results which they serve up for any vaguely related search so now it's impossible to add nuance to your searches as they all get lumped into the same bucket.

If your company is not asking you to do the things that are most likely to make it money then I'd worry about your long-term job security. In my experience, most employees lack the context to understand what the best way to make their company money is. Therefore this argument seems flawed. However, if you have a better way to make money for your company than your boss does, then you should be promoted. A good company listens to bottom-up feedback and adjusts its approach accordingly.

My Google Assistant is so bad now it can't even set an alarm which was the only thing I found it could actually do. It's bamboozling how bad they are.

Two scenarios:

- they're trying to prevent proprietary knowledge leaking;

- they're trying to prevent dodgy shit they're doing leaking.

My guess is the second as it's highly unlikely that they have any proprietary knowledge that would give their competitors a significant advantage to know. Most large companies' strategies are fairly obvious from the outside and proprietary knowledge leaks through employees moving. It's the dodgy shit they're doing that can damage them.

I think this is missing half the story. Niching down is a great way to differentiate in a crowded market and thus get started in the face of competition. The second part is to then expand our of that niche into adjacent areas, using the momentum you've gained through your differentiation. That's where the excitement and possibilities come. He's missed the second part of the journey.

GitLab task planning is so terrible we had to give up on it and move to Linear. It's like night and day. I can create ten tickets in Linear in the time it takes me to open the right GitLab board. The good thing is that this has probably saved us money as now fewer people need GitLab licenses.

There are several differences between LLMs and encyclopedias - LLMs don't need specific text about specific events in Chinese. They can be trained in text in English and produce answers in Chinese via translation. - LLMs probably need much more text to corrupt their veracity -- yet to be tested properly - LLMs are more obfuscated in their purpose. If I'm sharing a text about a specific event then that is pretty damning, but if an LLM can answer questions about an event then you can't prove that was why I was sharing it. In fact it would be pretty hard to exclude information about specific events from an LLM

Make me think the next version of GPT will learn all the rubbish prompts people use and rewrite them to be more effective. The next version of RLHF. I guess they're already using ChatGPT data to do this like Google uses researches to improve the initial results.

These new models could be the downfall of authoritarian regimes. These regimes rely on suppressing access to information but soon we'll have an LLM that can answer any questions about the world and fit on a USB stick that can be passed from person to person. No more great firewall of China.

Depends on the company. Conscious Capitalism and the Infinite Game are great books on why that's not the best approach.

I don't pay $200,000 a year for a car. If I were going to then I'd like to test drive it for 5-8 hours. An ideal test drive should help both sides understand what the future relationship will be like. That saves time and disappointment on both sides. You'd like to date someone before you marry them wouldn't you. I would always be happy to pay for a test drive though. If you're not willing to offer a test drive then you'll also probably miss out on some great relationships in your life.

They probably mean that if you test it on common NLP benchmarks it performs better.

And like everything, the price will drop rapidly as the models become smaller through advances. There's already sparseGPT.