Looks like the most notable this week are both from Lex Friedman's AI Podcast. Very glad that Alex turned me onto the Joe Rogan episode with him. Will probably have to go through the history of these. They are just wonderful. I think they show the best perspective on technology that I've found so far, mixed in with some extremely deep questions and fascinating topics. Only good things to say really.
Fucking hell this was one of the best podcasts I've listened to.
A lot of this will be notes from the podcast to best of my memory. I wish I had better timestamps when these thoughts occurred, but again this is the goal.
The first hour was particularly intriguing to me. It focused a lot on the things I'm most interested in. Specifically a few topics. All of them are interrelated so it's hard to pin thoughts to specific ones.
I think Jordan is right when talking about Facebook being creepy. You just see ads popping up about things you were talking about. It's just a black box and to see that is weird. Like you don't know what's being collected about you or when it's happening. You also don't know the relationship between that piece of data and who is using it and for what purpose. That's also scary.
I would want to be able to know these things about my data certainly. I also would want to be able to explicitly give out permission. No I don't want you listening to everything I say. I might have that data on myself (and want it I made the choice), but you shouldn't have it by default. If I do give it out, I want to know how that will benefit me. Why that data is being used or requested. How do you stop malicious intent? So many questions and interesting things going on in privacy.
There was a good amount of talk about data markets. One that was brought up (that I also love as an example) is music. Specifically how the distribution of money coming down for their creation is not very equal to the distribution of listening. Noted, is that Spotify wants to help in this, but they are still taking a large chunk. Also what if I as a listener want to give more to this artist than the average person. Noted in the podcast is the idea of metrics and knowing where to have a concert or other kind of events. Where to target merch etc. Allows artists to be closer to their fans. This is good for them. I've noted previously that I would love to be able to see the evolution of a piece of art/music too. I think I would have greater appreciation of it if I can see how it evolved and be able to explore its evolution. I think that could elevate the art immesely. I think this applies to all kinds of domains too. Also it could give artists such granular control in charging people for their music or provide new and different ways to make money of their creation. I bet you could charge wealthier people more. How do you get to know this information? Is this a good system? Does this equalize society? Or does it bring greater rifts? So many implications of a data market. Again it needs a lot of people to help out in creating it.
Data is only becoming more and more prevalant and valuable with the growth of IoT/AR/VR systems. These systems will fundamentally impact how we as average people interact with data, and I think pepole will start to be concerned very quickly once they understand how broken the system is. I don't think people will stay ignorant of the systems built around them and what they are doing. Even if they are, the people that aren't should be building better systems. Regardless I think they will figure it out because the data will be so valuable, so who would want to give it up for free? I imagine my data will be so valuable that I could sell it and earn a good amount of money from it just by living. I bet I could even charge more for it than if Facebook owned it. There has to be a data market to support this however, as the data in an of itself may not be that valuable. This raises all kinds of questions and has huge implications economically and socially so really need more intense thought with lots of people.
There needs to be a separation of data for sure and the current infrastructure is not built to handle this. A lot more data will be given up. Take health data for example, if I am embedded with a sensor in me is that going to go up to a cloud system of some random company providing the sensor? I need to know that it won't.
I also think that I don't know if I want a company in control of this kind of data system. It seems like a problem. I think such a system or platform can make money, but it should be like some sort of tax for using the system. To perpetuate it, and also provide money to improve other systems. Something that is self sustaining and governed. Governed by the people since they are the ones getting taxed. It is a global system and should have global participation. We should allocate money outside of the traditional government to things we feel are important. By using this system hopefully you feel that your money is going somewhere good. Maybe you can personally allocate it. Just a thought.
Personal knowledge systems/recommender systems
I guess more what was mentioned in the podcast itself was next generation recommender systems. Ones that take into greater account of your data and what may be interesting to you. More than just knowing for example I am going to "x" place what will be interesting. More I am going to "x" place and here are all the things interesting to me, please help me figure out what to do while I'm here. Could even help employ people and help them connect on deeper levels. Quite interesting concept and you can definitely see how this could be some kind of market.
Another interesting thing was the note on being able to control the the information I want and when. Jordan notes that ad's shouldn't go away, that'd be bad for everyone. I agree for the most part, as there is a lot of nuance to this specific topic (as with any). But on the main topic, I do want to control what I am seeing to some extent. In different contexts I want different information for sure, and I would love to be able to specify that to the system. Also I might want ads about stuff that my friends have or something. Know their personal reviews of what they are using. Explicitly or implicitly (well this can be scary too). But this level of control could be scary too. Depending on the implementation of such a thing it can lead to even more insular, and closed spheres than we have today. This kind of thing has to be done very carefully to ensure that we don't get so devisive.
Responsible Context Aware Data Systems
I'm slightly botching the quote, but here's the best I can do.
Around 13:30 onwards, Jordan is talking about how current systems, Google, Amazon, Uber:
"are slightly broken because there's not an engineering field that takes economic value in the context of data and planetary scale, and worries about all the externalities, the privacy [and so on]."
I don't know what to say or where to begin. This is the most compact sentence that sums up exactly how I feel about these systems too, and what needs to be considered for future systems. If you continue to follow my writings and creations you will probably find this is exactly the problem I am most interested in. I am not sure how else to state it succinctly.
The problem (and greatness) with this field described is that it is so encompassing. In order to change the systems for the better we must think of them fundamentally differently. They should be thought of from all perspectives from all fields of science and humanities and more. We must work together and integrate as many ideas and perspectives into such a field.
I think from the end of the podcast it is called something like Human Centric Engineering and Design. I think in some sense it is similar to human computer interaction, but HCI is a little more specific. This is a bit more broad and encompasses how other systems can be best engineered for humans. Computers are one major component of this system, but there are others as well. They all play into each other and require a lot of thought.
Not mentioned in the podcast but very related would be Personal Knowledge Graphs. This would be like my collection of knowledge, how I got there etc. People I know, all kinds of personal information. I imagine it would connect with other's graphs to speed up commication and make it more effective. It should link into the global knowledge graph as well. But there should be distinct separation between them. Like the global graph shouldnt have access to my personal graph unless its explicity provided. How to design this is a great question.
One other notable topic was surrounding progression of science/engineering with ideas or personalities. (19:00 according to the site).
The advice at the end of the podcast touched me and I will be reaching out to Jordan. I don't know what I'm doing and I am looking for mentorship. I've been thinking about grad school and a PhD, but my thoughts are scattered. If I were to do a PhD I would want to learn how to those scattered thoughts as consice as Jordan's quote above. It was beautiful. I am looking to in that cooperative kind of field. I am looking to learn and help others learn as well.
Just a quick note here really. I think the nice thing about Simon is the inspiration he brings to the table.
I had a thought which came almost directly from the podcast and I'd like to adopt it and attribute it. Unfortunately I don't have the timestamp
"I haven’t done much yet. It’s not about what I’ve done, but more about what we will do. I will help drive humanity forward"
Also shared this with Jon, he said he enjoyed.
This is why I hate the current data model.
I guess the data model is this one way linking thing. I'm telling you this is all a freaking graph. I don't see how it's not. probably some weighting to the graph as well. The graph is going to be huge in the project.
Roam does a better job, but I can't make it pubic and I don't think it wouldn't be easily browsable anyway.
That is I have notes from here which really integrate into other things that I've written, but not all deserve their own section. All this data is interconnected and I can't easily represent it on this wiki. Such a pita.
This is another goal of the project.
^ This sentence has so much data in it. That paragraph should be added to the goals of the project. But nope can't do that easily. Fucking hell