Transcript of Latest Public Talk by Dr. Richard M. Stallman (RMS), Delivered Last Month at Web3 Summit 2024 Berlin
Summary and video: [Video] Richard Stallman's New Talk in Germany Covers What Free Software Means, Why LLMs are "Bullshit", and Lots More (Web3 Summit 2024 Berlin)
Video only (WebM): Closing Keynote Day 3 - Dr. Richard Stallman - Web3 Summit 2024 Berlin
Click below to play:
Full transcript, including questions and answers (Closing Keynote Day 3 - Dr. Richard Stallman - Web3 Summit 2024 Berlin)
RMS: I'd like to point out that Free Software is a morally stronger idea and Open Source was an attempt later to co-opt the ... work of the community and de-politicize the philosophy. So that's why I have never supported Open Source in any way. I still support the Free Software Movement, that software should respect users' freedom. So, first of all, I'd like to say more about the difference between Free Software and open source because it's a topic of great confusion. I founded the Free Software movement in 1983 with the announcement. The term Open[01:00]
Source was coined fourteen years later in 1998 when Free Software was becoming widely used and starting to be something people knew about. But not everyone who worked on or used or promoted Free Software agreed with the philosophy of freedom behind it. And the people who didn't agree wanted to get out of connection with it by and many of them were working for businesses or with businesses that didn't care about freedom at all. So, they found a new term, Open Source, which they defined differently but it overlapped a lot. Most Free programs are Open Source. At the time it seemed all Free programs were open source, and most open source
[02:00]
programs were Free. So, it is easy to get confused and think of these two as identical, but they are not. But the biggest difference is that the term Open Source has never had any implications about right and wrong. It was, that idea was launched that way by people who didn't see it as a matter of right and wrong. So that's why I decided I would not start using that term. They asked me to and I said, "no". Because Freedom, in my view, is what it's all about and I want the idea of Freedom to reach everybody who comes in contact with it. So what this means to you is that you have a choice. If you talk about Free/Libre software you will be saying I'm talking about Freedom. Freedom is what this is about.
[03:00]
And if you use that term consistently you will spread that message. If you use the term Open Source you'll spread a different message. Well, what you say is up to you. But if you care about Freedom, please find opportunities to show that you care because we are competing with a different idea which has a lot of businesses promoting it. In order for use to keep the idea of Free Software alive, it helps if we work at it. So that's why you'll never hear me use the term Open Source. I mention it in order to criticize it, but I don't describe software that way. Now, there's a new confusion which has sprung up in the last couple of
[04:00]
years. There are various kinds of artificial intelligence programs that within specific domains can understand a problem and can understand how to get the correct answer for that kind of problem at least usually as often as humans can do it or more but one kind of program which is not intelligence is the Large Language Models because intelligence implies understanding and those programs generate output but they have no idea what the output means. They're thinking of word usage only. And so we shouldn't be surprised that they generate statements that are false
[05:00]
very often or even statements that are almost nonsensical. They're grammatical but they don't mean anything. They present imaginary fictitious events as if they were real, and yet they are being called artificial intelligence and most people on seeing that assume that the output of these programs can be believed. But it can't be. They have no idea of what's true. They don't understand the statements they generate an so we shouldn't call them A I and I never do. Sometimes I call them bullshit generators. Bullshit is defined as generating statements, producing statements with indifference to their truth or falsehood.
[06:00]
Of course if you can't understand truth and falsehood, you can't be anything but indifferent to it. And that's what those programs are like. There are also humans that output bullshit who are presumably capable of understanding whether they're true or not but don't care. For instance, Trump. [applause]
But I suppose as a human being, he would be capable of caring about the truth of his statements if it ever occurred to him to care. You know Trump has no heart but he still needs a defibrillator. But a program that can't have any idea of what is true certainly can't care. So we know that those bullshit
[07:00]
generators are not intelligence they can't understand. So, moving on from that, one thing we can see is that a web site should never use a bullshit generator to do any job that depends on accuracy or validity or truth, because it's going to go wrong and more often than you might think. So it's a very bad thing to change a web site to be so-called smarter by having it pass what you tell it through a bullshit generator or having it pass what other people have published through a bullshit generator and giving you a summary that might be total nonsense.
[08:00]
It'll be written in good English though. Now when you are doing your computing you must not entrust that to somebody else's server because users including you should have control over their own computing but you can never have control over what somebody else's server does because somebody else installs software in that computer and configures it and thus decides what computing it is going to do. So it may be useful to talk with somebody else's server as a way of communicating
[09:00]
with her with somebody else. Servers are fine for that. If I want to publish something that I say and you want to see what I said connecting to my server stallman.org is a fine way to do it. And running a server for that purpose is a fine way of conveying the information to those who are interested. The GNU Project has its own things to say different to my personal views so it has a different server gnu.org and again it is fine for the GNU Project to use a server as a way of enabling you to see what the GNU Project has to say if you're interested. And the Free Software Foundation has another server fsf.org which gives other information about a related part of the issue of Free
[10:00]
Software. But when it comes to doing your computer you shouldn't entrust anyone else's server, not stallman.org, not gnu.org, not fsf.org, not google.com, or apple.com, or basically any one else that sets up a server that is not you. And that person or organization's server is not your computer, it's not under your control, and so you shouldn't depend on that server to do your computing. Which means that a very wide spread architecture that is quite commonly recommended by the people who don't think about this issue is a bad way to do computing.
[11:00]
It might be perfectly fine to use a certain Free program by running it on your computer. That way you can change it if you want. But to have the same program running on somebody else's server, and you just connect to that server, well, you can't change the version of that program that's running on that server because that server is not yours. So it's not just which program you're talking to, it's who's running it that matters in terms of your freedom. For your freedom it should be you running that free program, so you can modify it. If it's running in my server, well, that's my copy and I can modify it but you can't. At least you can't put the modifications into the copy running on my server. That's not good for your freedom.
[12:00]
Now one very visible aspect of this danger is that there are a lot of programs which people think of as sort of like Free or Open Source which impose limits on what users can do with them. Now that can't be the case if you install a Free program and run it. A Free program respects Freedom Zero, which is the freedom to run the program however you wish. But there are programs such as Ollama or is it llama? I don't know. But the point is it's not Free Software because in its license it says that you have to agree to a contract in order to use it and that contract has a long list of
[13:00]
things that you're committing never to use it for. It makes no difference whether those things are bad things or ok things the point is that companies should never have that kind of power over society. What would happen if society were to pick up the use of this quote "Open Source" unquote program and everybody was expected to agree to the license's restrictions on use that are included in that Ollama license. It wasn't easy for me to get to see it. You have to run non-Free software in order to download a copy of the license. I had someone send it to me. I hope
[14:00]
that person won't get sued for copyright infringement for sending me a copy of the license. The point is that we may want to place restrictions on things people are allowed to do with that program but if those restrictions are justified then they're not justified only for one program. In order for them to be justified there has got to be a powerful reason that would apply to any program that can do those same things and it should be adopted by governments democratically legislated not simply proclaimed by a business and imposed on society. So if we want, and I think we should have, some laws about what we are allowed to do
[15:00]
with certain computer programs we should make them general enough that they apply regardless of which program you might use and we should not let businesses decide what they say. We must not farm out legislation to powerful businesses they have no right to have that power. Consider for instance fraud. Fraud is wrong. It's wrong whether it's carried out using a pen and paper or with a text editor, with an e-mail client, whatever. And in fact, it's prohibited regardless of those details because the crime is fraud. The crime is
[16:00]
not misuse of a text editor or misuse of e-mail or misuse of a pen. And that's the way it should be. It means that we don't need to update the law against fraud to apply specifically to all the programs you might use to commit fraud with. They're all covered because those details are irrelevant to that law and that law was not decided by a company. It's not imposed on us by a company that everyone feels is inevitable to be a customer of. We should do the same thing any abuse of digital technology which is bad enough that it needs this kind of correction to be done by laws that are
[17:00]
general not associated with the release of any particular program. And not a matter for the developer of the program to choose. The developer should not have the power to impose such restrictions at all. Only governments who get the power to legislate should be allowed to do that because with governments we can have democracy but businesses will never be governed by our votes.
Now, another problem we have with the Web as it is today is that it has become common for web sites to send software to run in users' browsers. The World Wide Web originally
[18:00]
operated by transmitting declarative data to a browser and the browser would then, I can't remember the word, render that data as some sort of picture on the screen. And the browser would do this according to code that runs in the users' machines so once there was a free browser users could control how that data would be rendered. The web was designed to let users control how that data would be rendered but businesses didn't like that. Businesses wanted to be able to control how their pages would look to users. And they redesigned things to give
[19:00]
the business more and more say in how its pages would look when rendered by a user. Well to some extent it's optional, as long as it was only CSS you could turn it off, you could delete it, various things ultimately you were not under the business' control. But with Javascript it's much worse. Javascript as it's normally used is a program sent by a business or other organization to your computer to run there and practically speaking it is very hard to be able to change it. There are ways. It's a kludge though. And you could hardly ever have the time to study the sometimes very
[20:00]
large programs that so many different web sites send you. You may look at that site only for a few minutes and then you're looking at something else no one can feasibly look at these and there's no easy way to distribute to people recommended changes and you don't normally get the real source code either. You get obfuscated code or you get compiled code in the form of assembler language or just numbers. This is no good. We need to go back to a system more like the original web where what you get sent is declarative data, which is then realized as
[21:00]
standard kinds of behavior so that normally web sites can work without any specific Javascript code. This means we have to come up with general constructs that can be combined to get a lot of useful results, something like the 1990s windowing toolkits. You combine those, you can get the behavior you want, and it will display reasonably fast but always using code that is part of the system you installed and so we can build up a suitable library of general constructs with various parameters to be set for each one each time it is used and out of those you could build your web page.
[22:00]
It would do the specific things it is supposed to do but the thousands of web pages you would look at would be using a finite list of general constructs that can be developed and maintained as Free Software. And who's community verification and maintenance would give you a reason to trust it. In this way, we would have much less malware in the world's web sites.
Now the decentralized web, if it is to be for everyone, must make web pages accessible to everyone. Including stalwarts like me who
[23:00]
don't run non-Free Javascript code. So, one thing you've gotta do for a world like this that everyone can participate in is to make sure your web sites are accessible without running non-Free Javascript. And ideally accessible without running non-standard constructs so that your system could include implementations of it.
When I first looked at your web site for this event, I couldn't see most of it because it wouldn't display much without running non-Free Javascript. I think that that was fixed because later on I could see most of the site. This should be a
[24:00]
criterion we make sure we fully follow and you have to watch out for making a web site use another web site to do part of the job because you could make all of your Javascript code Free and forget to check whether the other web site's code is Free and the solution for this is to try browsing the site LibreJS enabled. Which will tell you about and block non-Free Javascript. That way you can make sure you've made your web site fully accessible to stallwarts, freedom defending stallwarts.
[25:00]
Now, decentralization must include decentralization of what software is running. The July 19th Crowdstrike disaster shows what happens when lots and lots of web sites are running the same piece of software that gives special access to the same people. They might all go down the same day. And who can fix them? There are only a few people who are even authorized to try to fix them. We better learn our lesson from that as a world and not get into that sort of power of some few.
[26:00]
We have to recognize that letting a few have such power over our computing activities is not only threatening to be oppressive but threatening to wreck things completely.
Anti-social media platforms all put a lot of effort into their recommendation engines. And these are examples of programs that I consider to be truly artificial intelligence because they do understand the field that they are supposed to understand, they're designed to understand, and that is which kinds of recommendations will enable the company to make more money. That's what the companies want, and the programs can figure that out pretty well.
[27:00]
For instance, anything that is upsetting, controversial, makes some people angry, is likely to keep people on the site and that's what the company wants. More time on the site means more ads seen and so on. And yet that pattern turns out to be very harmful for society. It has boosted right wing extremists and puts democracy in danger. I have a recommendation for how to solve that. The platforms themselves should not be allowed to have recommendation engines. This is something like anti-trust law. No, you shouldn't have one company able to let people post things and
[28:00]
also choose what to recommend what people watch. Instead the recommendatio should have to be separate. And the two should operate at arm's length through a publicly specified interface so that anyone else can make a recommendation engine. Anyone could, there'll be competing recommendation engines you could use with the same platform or perhaps with various platforms and that way yes there will surely be a right wing extremist recommendation engine but it won't be the only one and you won't be shoved straight into that whether you like it or not. Some US states are considering passing laws
[29:00]
that platforms can't direct their recommendation engines to quote children unquote which means people it's not clear what age. They're not all actually children but the problem with that is these recommendations are dangerous for adults too. So you're not really solving the whole problem if you try to protect only children. But I'm not saying ban recommendation engines, I'm saying disconnect them from platforms so platforms can't dump recommendation engines on poor users, or on any user if the user doesn't choose it.
[30:00]
One other point to mention, GitHub is not a good site to use. Most operations on GitHub require running non-Free software. I can't get much information out of the GitHub URL. I mightg be able to download and clone a repository if I wanted to study in that much detail which I never do. I never have time for that but to get the general information about the program, like what does it do, what is its license, what does it depend on, that's not easy to get. That's a very bad thing. In
[31:00]
gnu.org/software/repo-criteria-evaluation.html you can see our criteria for judging a repository site and our evaluations of various repository sites including GitHub. GitHub gets an F, and you can see exactly why. So please don't put your work on GitHub. Please put it somewhere else on a site that matches up to the criteria for a good and Freedom respecting repository .
What about payments? I've never used cryptocurrency. There were things I found disappointing and worrisome
[32:00]
when I learned about BitCoin. And it's not clear to me that others are much better. One thing is that BitCoin tries to make the payer and payee, both of them, anonymous. Now it may not do a very good job of them. Basically it's the wallet doesn't have to be identified with anyone but since all the transactions can be tracked, the money can be tracked around from one person to another and unless you are a miner how did you get it? That's one reason why I prefer GNU Taler. GNU Taler is an anonymous payment system developed in the GNU Project. You can get more information at
[33:00]
taler.net It's not a currency. It's for payments denominated in currencies that people use. So if it were implemented here in Germany it would probably be trading with Euros. But it's not an investment vehicle. It's not meant for currency speculation because it's not a currency. If you get say a hundred Euros in Taler tokens well the only fluctuation in their value will be the fluctuation of the Euro which is much less than the fluctuation of a cryptocurrency. I don't want to do currency speculation myself at all. I'm not a fan of risk. I don't gamble. Now with Taler the payer is anonymous
[34:00]
but the payee is always identified, which means that Taler does not help millionaires hide lots of money from taxation. The world has a tremendous problem with wealth that is hidden and cannot be taxed. It's part of the way that billionaires have been transferring more and more of the world's wealth to them leaving less and less for everyone else. And this change is on the order of twenty percent of the world's wealth. It's an enormous change that impoverishes people who are not rich but even worse it gives the rich people the power of oligarchy, the power to buy governments
[35:00]
and that threatens democracy. That threatens the rights of all of us but if we insist on payment systems that don't permit the hiding of large amounts of wealth, that problem will get less instead of more. Taler is not designed for paying a million dollars or a million Euros. It's designed for buying things every day.
So, I went to an event in October and saw somebody wearing a shirt that said tax me if you can. His reason to advocate cryptocurrencies was essentially to try to
[36:00]
exclude his money from contributing a share to the needs of having a good society. Their states have very important responsibilities and in order to carry those out they need taxes. Letting the rich get out of the taxes and dump those on the non-rich leads to disaster. So I'm happy that GNU Taler will not have that flaw.
So that's that.
Now I'm ready to tell you one final word, which is please join
[37:00]
the Free Software Foundation. The FSF needs your support, and if you join at fsf.org we've gone to a lot of work to make sure that you can do that payment without running any non-Free Javascript code. It's important for us to practice what we preach. So please join or donate, and after the questions I have a gnu to auction but since the funkhaus does not allow selling inside I'm going to invite people to go outdoors with me and I will have the auction somewhere else not far away. But in the mean time. Walking distance basically. But right now it's
[38:00]
time for questions and I would like the organizers to tell me they've handed out pieces of paper to write the questions on.
MC: Yes, we have the questions on paper. Would you like me to read them to you?
RMS: No. No. No. No. No. No. Hand me the pieces of paper. I have hearing problems. I can read better than I can hear. Bring the papers to me and let me read them.
MC: Coming right up.
RMS: And please keep going around. Every time there are more papers, collect some and bring them to me. We can have them in a queue. Thank you. Keep seeing if people may be writing more questions now.
[39:00]
Your thoughts about Web3? Are cryptographic guarantees sufficient to trust personal information? Doesn't it contradict with Free Software? Well, there's no contradiction in principle between free software and non-published personal information. First of all, the personal data is normally not software and in the Free Software Movement we don't say that every piece of information has to be published. In general we are in favor of privacy for our personal information and your personal information. We don't say even that every program must be
[40:00]
released. You are free to write programs and never show them to anybody. We never said you shouldn't. In fact, Free licenses, part of being Free with freedom one is that you can make a modification and never publish the modified version or the diff or anything. You're never required to publish Sany version. What Freedom three says is that you have the freedom to publish your modifications or your modified version, but you don't have to. You can get a copy of GNU Emacs and make edits to serve your convenience and you don't have to publish those. Now Copyleft
[41:00]
says that if you do publish those, and others download that, or get copies of it, they also must likewise get the four freedoms. But you don't have to give a copy to any particular person. You don't ever have to offer copies to anybody. Freedom three doesn't say you must publish your modified version. It only says you are free to do so, if you choose. So there is no even hint of conflict between Free Software and Copyleft and privacy for things that are personal. If you write a program for your personal use, by all means, keep it private if that helps you. Now if it were a program of some general usefulness
[42:00]
such that having it would be a help to a lot of people then I would urge you make the program more general, make the things that you personally like, be controlled by parameters which you could specify one way and I could specify a different way and then release your program contribute it to the community. But it's never required that you publish a program or distribute it to anybody at all. So I think that means there is no actual contradiction between them.
Could you think about how block chain tech can help defend human rights related to Free Software.
[43:00]
I am not sure. Bruce Schneier a few years ago said that in his view block chain didn't make possible anything that couldn't be done without it. I'm not an expert in that field and I don't know, because my concern is not about software like that it's about making sure the software we use is Free. So I saw that the various cryptocurrencies people were talking about were implemented with Free Software and I said, ok that's good on that score it's being handled the right way. And I didn't look at it any further because that is the aspect of
[44:00]
things that I focus on.
Is bullshit ok if people agree on it, with a consensus system. I'm not completely sure what a consensus system is. Bullshit may be ok as humor and dissociated press is fun. So maybe so. The point is that bullshit becomes dangerous when it's presented as truth. So perhaps if every bullshit generator's output was interspersed with lines saying, this is output from a
[45:00]
bullshit generator don't assume anything in it is veritical [sic] then it might be harmless. I'm not sure.
Have you heard of the term quote fair source unquote? I don't know. I don't think I have. And the functional source license? I'm not sure what that is. Sorry. Maybe you could look at gnu.org/licenses/license-list.html and see if those licenses are listed there. If they are, you could maybe tell us what they say or if they're not maybe you could e-mail me those licenses so
[46:00]
we at the FSF could think about them and judge whether they are free licenses. Can't we just call it Free and Open Software? Well, if that's what you want to say, I can't stop you but I think it's misleading. First of all do you mean the set that includes free programs and open source programs. Or does it mean programs that are at once both free and open source. You can see one of these is a conjunction one of these is a disjunction. Free or you could take that as meaning Free is synonymous with open source but they are not synonymous. They're very different. I recommend reading gnu.org/philosophy/open-source-misses-the-point.en.html
[47:00]
which explains thoroughly every aspect of the relationship between Free Software and Open Source Software.
How much do you know about Web3? Not a tremendous amount, that's not my field.
Have you ever written a smart contract and if no why? Well first of all I've never used Ethereum and as far as I know that's the only way smart contracts get implemented. I have not had a hunger to do it because the idea makes me worry. In society contract
[48:00]
can be dissolved, they can be voided, and this is sometimes a very good thing. For instance it makes possible bankruptcy that means that if you sign up for a contract that requires that you work for the rest of your life it can be declared invalid because legally one can't bind oneself for the rest of one's life like that. I am worried about whether some sort of computerized contracts will eliminate protections and freedoms such as bankruptcy law. I expect that a lot of businesses would jump at the chance to set up contracts and invite people to sign them which will trap them forever and they can't get out.
[49:00]
Some software is advertised as open but not open source.
How do you apply the Free Software definition for AI models. Well we're thinking about that question right now. So I'm not sure. It has to be in principle the same freedoms. So how do you apply it to some sort of computer system where part of in addition to programs there are also collections of data which were not written by anyone but were
[50:00]
produced by operation of the system, which is what we mean by training. And the result is data which is useful for achieving a certain result when the program operates on it but was never designed by humans so there is no source code or anything comparable to it. Well, one way of looking at it is, can you change that data. If it were a program with source code you could change the source code and so yes you could change the thing and how it would behave. Well for machine learning control data, it would have to give you a way to change it. One possible way is continuing the training, incremental
[51:00]
training. You could use that to change how the system would behave. But it's a fact that since these people can't understand the meaning of the data in that collection there's no way to try to get a precise result. You train one of these recognition systems and later you'll get surprising behavior some of the time, and that may be a general rule. I'm not an expert in that field. But if
[52:00]
the most you could ever expect is a sort of approximate correctness an approximate fit with the specifications you want to achieve, well maybe it's good enough if incremental training gives you that much. If it's the most you could ever hope for in that field, it'd better be enough or else you don't want to do the thing that way.
How to make people care? I think that's care. You can't, as far as I know. You can't make people feel this or that. People's own minds contribute to what they care about to their choice about what they care about. What I find is that. It's good to present the issue in ways that relate to things that people care about in other areas of life. And in the 14-minute video you can see where I got to
[53:00]
following that approach.
Should I have the freedom for creating, distributing, and supporting proprietary software? Well, first of all, I think you shouldn't do those things but that doesn't mean you should be prohibited to do them. I tend to think that if there is something people want, prohibiting it is often not a very good solution. Look what happens from prohibiting dangerous drugs. On the other hand, maybe you can figure out why you see a lot of people trying to use dangerous drugs maybe you can see what leads them to go in that dangerous direction
[54:00]
and maybe make something else available that would be better than a drug maybe people wouldn't be drawn to those drugs. I don't know. The method of supplying drugs to addicts so as to wipe out the black market for drugs and then it would be harder for anyone who is not an addict to get her hands on some drug. That seems to work pretty well in countries that try it. So maybe a similar general approach could work with non-Free software as a way of discouraging it from being a successful business.
What did you learn from the attempt to cancel you? Well, I learned
[55:00]
first of all it's not clearly a thing of the past. I wish it were. There are still people who make it their business to try to stop me from getting invited to speak, and it's a slow process working back from that. However, I learned the ability to stay calm and not react by getting upset and instead think carefully about what I should do. I am much, much better at keeping my self control when I respond to antagonistic attitudes. In fact it was amusing. I got a hate phone call about that time and
[56:00]
someone I didn't know said, "you defended Epstein", I said I didn't and a couple of interchanges like that and I brought it into a reasonable conversation. I felt pretty good about that. We went on for a while talking about various questions of right and wrong and how I had tried to do the right thing in connection with them. And what I thought, because I certainly did not approve of Epstein and I appreciate that his actions were quite hurtful and I never said otherwise. So she was able to understand me too.
[57:00]
This question seems to be written in such a way I can't see it. Or maybe just a blank sheet of paper got in.
How or in what capacity does Web3 movement influence Free Software initiatives. Well, if you develop some Free Software, those are Free Software initiatives and you're influencing each other. Basically, nobody is in charge of Free Software, in general because every person or group contributing is free. That's the whole idea. Because of that I don't know how I could answer more than that but if you think it would be a good idea for me, the GNU Project, or the FSF to
[58:00]
recommend people hey please try to develop something to do this if you can, we can think about it and maybe do that. That might help out.
More and more software is produced it says using AI. I don't know whether this refers to things I would call AI or things I would call bullshit generators. As these apps have become so easily created they have become a commodity and copyright and licenses become irrelevant. I don't see that that's true. But it's so vague I couldn't tell.
How can people here donate to the cause of Freedom?
[59:00]
Well, the cause of freedom includes lots of political causes. There are many different freedoms, such as freedom of speech, the freedom to have a fair trial, that are tremendously important to all of us, not particularly related to software directly. There is freedom to have an abortion, which is not directly pertinent to me but it's very important cause and there are so many others. I can't give you an answer to a question that broad. But if you're talking about computing freedom and software freedom, well most of these issues there is nothing but the FSF that campaigns for it. At least that's
[01:00:00]
the the big visible organization but there are also lots of local organizations that promote software freedom in the area where they're at and there are a lot of those and you can join those. You can also donate to the Free Software Foundation. Please do. We need your support. You can also become an activist yourself, but donating to the FSF you can do through fsf.org.
How do you see freedom of transaction in your terms? I'm sorry, I'm not sure what that means. I can imagine various possible meanings for it. Maybe write another question as a follow on.
What would you say to people who think AI is
[01:01:00]
sentient and equal to us? Well, the real AI systems that I've heard about operate in very narrow domains. A program that can look at an image of some cells and report whether it's cancerous at least as accurately as a human pathologist, that can be very useful. But it's in now way comparable to the mental breadth of a human mind. And no one would ever think it was. I suspect that the person who wrote this question was thinking of bullshit generators. And there, well there are a lot of things I can say to them. I saw a wonderful
[01:02:00]
lecture by professor Zittrain of Harvard showing just how little bullshit generators actually understand of what they output and I think it got the message across very clearly. But I don't know whether he's published that. But I think that the first step is stop calling them AI. That's why I put so much emphasis on what we should call them. After all, what we call them doesn't change what they are. If you call it a bullshit generator or you call it AI it will still be the same system, whichever one it happens to be it'll still be the same. Whichever system you're talking about, it'll still be the same whether you call it AI or a bullshit generator but what people expect from it will be different.
[01:03:00]
The practice of calling these things intelligence and repeating this many times a day to people leads most people unthinkingly to assume that that's true. That's why it makes a difference which one we call them. That's why it's not enough to say yeah they call them AI but it's not true and then go on calling them AI. If we want to change the direction of the public of what to think, we have to be heard on this whenever people talk with us about this things and call it AI. We've got to be there and say, that's a misnomer they are not anything like what we think.
I suspect GitHub is
[01:04:00]
the tip of the iceberg, from here freedom is perhaps too expensive as it will require full stack chain. I don't know what you mean by a stack in this context. But if you don't want to host on GitHub there are better repository sites more respecting of freedom that are listed in that same page gnu.org/software/repo-criteria-evaluation.html sorry I don't remember if there is an 'S'
Outside of tech, what do you like to do? What are your hobbies? Well, I used to have some hobbies which I found inevitably I would give up. I used to love
[01:05:00]
international folk dancing but an injury meant I couldn't do it any more or other kinds of dancing that I used to enjoy. I used to enjoy playing music on a recorder but ten years ago or so I found that even when I practiced up I couldn't play tunes and have them come out right that even the same tunes that I had loved and learned how to play. I don't know why that is. I gave up. I still enjoy eating delicious food. It's not as easy as it was because I don't dare eat indoors with other people around I can catch COVID that way and I could end up dead.
[01:06:00]
So I can only eat with people in outdoor restaurants at outdoor tables and that limits what's available. Or if they come to my house I can turn on the HEPA filter that I've got for this purpose. Ok, sometimes one or the other will work but it's a lot more constraining than it had been.
Bullshit generators are objectively helping users with efficiency. Are they? How does the Free Software Movement reconcile with this fact? Is it a fact? I don't know that it's helping them. What I hear is that it's
[01:07:00]
often making users frustrated but I have no personal experience of this since you have to run non-free software also and to generally talk to them. And since they are running on someone else's server, I wouldn't be willing to actually use them. To actually use someone else's server to do computing for real use rather than just playing it's what we call service as a software substitute and it's a way of doing your computing and losing your freedom in it. So I won't do it. So I have no experience of my own with using those systems but I read plenty from people who do. Which
[01:08:00]
casts doubt on whether they're really helping people in a useful way, saying if I wanted to try to write something and make it sound like good English without actually having to write it, it might save me some time but I'd have to read it very carefully to make sure it hadn't made any mistakes. Semantic mistakes that is.
Transcriber's disclaimer: "This is an unofficial quick-and-dirty transcription."
"He [RMS] had some very good points," he said, "but due to the length of the session they get lost."
We'll refrain from making any further comments and instead refer to the comments made yesterday. There is an outline of the topics there.