ASI Safety Lab

Things I wish my computer can do for me…

I have several computers in my home. Some of them are switched off, or they are running and doing nothing. Even if I work on them or use them for seeing videos or any other seemingly heavy-duty stuff, they are basically doing nothing. I have an (always visible) icon in the sys-tray to see the used memory and how little my multi-core PC and laptops are being used (see pic below). Usually, 5% of what the PC could do, sometimes 10 or 20% (that’s all). I know that my Notebook is slowing down to save energy — so it can speed up the CPU by a factor of 2 or 3. Right now, I have no use for this performance — but I want to use it.

In my blog post on “How we will interact with ASI”, I was calling my PCs and Notebook lazy, and I promised to give a rant on what my machine should do for me. Something was itching in me to get this post done much sooner than I thought. So, here I am.

I want to start by stating the obvious. I don’t want my machine being used by some external hackers — not with anyone that I am not granting explicit permission. I don’t want data intruders checking out data or my internal sensors. I don’t want to be spied on with my own videocams and/or microphones in my home. All that is mine — under no circumstances I would allow some malware, virus, spyware, or backdoor to my systems. My house, my rules. I value and enjoy my privacy. If that is guaranteed, then I can answer the following question.

So what can my home PCs do for me?

Actually, I want a complete inventory of what I have or what is currently in my environment. Everything: there are pictures/paintings on the wall; there were once flowers on a corner table; there are 6 cans of mushroom soup on the shelves, in case I spontaneously see them during my shopping and ask myself if I need more — I know I can get an answer. These things will move within the house, I want to know where they are and potentially where they were before. I want to know which books I have or magazines I read, and where they are. I want to know where my food is, how much is left — when I bought it and what kind of stuff I can make out of it. I want that my computer knows where I have left my keys, or which different T-Shirt I have and which one is currently in the laundry. Well, getting these kinds of information from video data and getting them organized by my PC will take a while — visual object recognition together with automatic detection of the environments and point of views (including different lighting conditions) and all making sense out of it (i.e. using context) that could take a while. Anything that would come in new, even a book or new boxes of food, doesn’t need to be explicitly introduced or shown to the system, my local computer system is relentless in trying to get the required info on its own using the different video feeds available in my home.

Actually, I can imagine having reliable video surveillance of all my rooms using my Home PC — but without being concerned about privacy violations, it knows when to pause and to delete images. Technically, that is already available, but without an accurate 3D model of the interior, I will be disturbed by false positives from sudden changes in e.g. the sun’s brightness (from clouds) or from my cat or dog running through the house. But then my system should help me to make me aware if the dog has done something that I should know, like biting in my leather couch — and yes in that case I want to be informed immediately.

I buy groceries regularly, why not having my PC tracking all my expenses in detail. And yes, I would like to live more healthily. Therefore, my system will track what I do, what I eat/drink, and how much. All this can be done automatically. My system would give an estimate of my calorie intake and my calorie use due to activities.

Some with knowledge of Iron Man would say, you want Jarvis (J.A.R.V.I.S) — actually, I would like to have something better than that — but Jarvis is a good starting point.

I would like to instruct my computer to search for options for a healthier life. At this point, we are already talking about something that is beyond tracking and accounting. It would require AI to help me to make healthier lifestyle choices — and this based on knowledge about me. I could use an online app for that, but it would be better for my system to choose between multiple suggestions and to make a decision that would fit my style best. I would definitely appreciate it if my home system would sync and collaborate with my smartphone and its apps. AI could also prepare itself for the situation to suggest some meals related to the wine I have in my cellar — and even educate me further on wines so that I become a better wine connoisseur.

My PC should also be able to follow the news; it has learned what I am interested in. Currently, companies expecting me to share these data with them. But my interests, my priorities are my private business, and I am very uncomfortable sharing that with some companies that intend to sell me out to some advertisers. Much better would be that my computer gives me (personalized of course) a mixture of different sources while pointing out differences in how different media are reporting on it. Because I have asked a certain question before, my AI is already being prepared to answer that question if I ask again.

During the time my PC is not assigned to tasks, software on my system could do some research on things that I have shown interest in before. I can have regular conversations with my Home-system on what I want to learn, what I am still not understanding, and it will find documents or videos that I need to see for a better understanding. It may even give me the info that others got confused about. This requires a certain level of AI and potentially even ASI, but this kind of interaction is in the range of what we can expect from AI (or to be more careful: that we can hope to get from AI). I hope we create open repositories of information in which human intelligence is assisting AI to organize human, cultural, and technical knowledge in a way so that humans can learn faster and/or correct misconceptions faster.

So far, I believe I have not mentioned anything that would make my home-computer systems really busy. Efficient algorithms will get even the initial 3D scan of my home/belongs done relatively quickly, the same applies to the identification of all objects within my home. Similar efficiency can expect from regular tasks like indoor surveillance or news preparations.

But there is much more: I have many online accounts. My entire online/usage history from these websites/apps should be made available for my PC. My home system is then trying to categorize and is getting summaries or even the full content to analyze. It may process video/audio in multiple runs because it is learning, and it may get more in each run. Ideally, my system should anonymously exchange learned info/context with other systems that are doing similar things in the background. At some point, while seeing a movie, I may get curious about where a scene (e.g. a beach) has been shot. My system can’t answer it, but that kind of question got stored and a few days later, I may see a list of possible locations with additional info – or I lost interest, but the AI learned that the real-world location of a scene might be of interest. The advantage of this kind of P2P learning is that the next users asking for this kind of information will receive from the system better, more accurate lists of possible locations.

Most of the software tasks I mentioned above are better suited to be run and analyzed by one of the billions of home computers. Even finding scene locations in videos would be a distributed process for which people who have had similar questions would donate their (unused) computer resources.

I am a curious guy, and I would love to consider myself a citizen scientist. I don’t have sufficient education in biology, but if asked to take a walk in a certain area of a forest and make a video recording, so that these data can be synced with some satellite measurements done on the same day, I would gladly do it. If asked, I would use a drone with a video cam and grant a scientist remote access. Processing all these videos in detail could require some time, but I would love to participate – and because it’s only machine resources, I can safely donating it to a cause that I like. The algorithms that I was asked to run could e.g. measure and calculate the different biomass in that environment even if it would take a few days. Using my PC for these purposes makes sense to me — keeping my PC lazy not — it is frustrating.

The question is who will write all this software. Initially, we human developers of course. And collectively, we human customers will create a lot of data when we help ASI in identifying things or correcting them. Actually, having humans helping AI is already a service from Amazon’s Mechanical Turk. And then developers will work on better tooling so that we can accelerate the delivery of results and/or providing common sense to things we can’t expect from a machine to determine yet. And if ASI is finally coming, then it would help us finish what we have started. And then: no worries, there are many other projects in which we need large amounts of computational resources (and I am not talking about cryptocurrency).

There are many things that people would prefer not to tell the outside. To name a few: details on someone’s physical or mental health. Test have shown that people would be more open and honest with a machine than with a fellow human. So, I understand the business reason why healthcare providers would prefer to offer services via centralized computer facilities — but this doesn’t need to happen within the first step of the patient-doctor contact. The initial intake and interview could be done by the local PC (I know that some companies are working on some online solution, but that is a business decision and not a decision based on technical reasons). The same applies when someone needs help with their mental health. A human professional would always be available, but who knows, some people would be more comfortable with an equivalent AI instead.

Additionally, many people are living in disaster zones. My computer system could figure out in advance how I could prepare most effectively. My system would not lead me into a sales talk, but it could give me a realistic picture together with probabilities. It could ask me about my priorities and could give an assessment on what kind of damage I can expect and what actions (if anything) could be done to mitigate that situation.

What makes this approach different.

Currently, Home-PCs are mainly used as an interactive tool. If I am not working on it, it doesn’t have to do anything because it’s usually done so fast. It may be used to play videos or music, but that’s it. Somehow, we can’t imagine that our PC can really get busy for us. I already mentioned a couple of tasks and I hope that software entrepreneurs should take a look into them. My mindset is: I don’t want to get busy, I would love to see my PC busy. And I would love to see that it would find out (sometimes) something surprising about me.

My PC should be busy (proactively) collecting data about me, my health, my fitness, the food I eat, what I do and don’t do enough – via IoT or “more conventionally” via video-cam. With my data, I can improve my life quality continuously.

There are a few fitness and health apps collecting data, to be stored in some external company servers of course. I want to pay for products that give me my data on my systems, the more, the merrier, and then I want non-proprietary tools to have these data analyzed. I may even share my data for the benefit of science, but to be sold out to another advertiser: no. In concluding this rant, I might ask about a closely related question: Why is it so difficult quantifying myself?

Privacy Protection

The main objection would come from “why I would accept the risk of being spied on”. Local video-cam and home PCs are vulnerable. All this would become attractive targets of outsiders to intrude on someone’s privacy. Therefore, today, I would not do anything of this kind. I deeply distrust my PCs and every IT equipment in my home and office. I distrust virus-scanner and firewalls. Without the technology I was suggesting in item 7 “Software Validation and Updating” and item 8 “Port Certificates in Firewalls” in my posting on “Standards in ASI Safety”, I wouldn’t do it. Both technologies are simple, and they wouldn’t affect me at all except that I won’t need to be concerned about Malware, Viruses, Spyware, Ransomware, Rootkits, or Backdoors anymore. Related to having external code and algorithms running on my system I would insist that it must run in a Java VM or in a VM-ware, that could potentially even be a precursor solution to item 11 “Imprints in Executables” as mentioned in my post “Standards in ASI Safety” in which the VM ware is being started automatically when an external app is running on the system. However, pre-ASI, and without having a machine with an ASI watchdog, Imprints in Executable are a non-starter anyway.

I believe I have just scratched the surface on what home computers can do. It can be much better than any online service. Mainly because my local system could protect and defend my privacy and autonomy. Advertising-funded online services can’t be trusted. Many paid online services are (so far) too rigid in how they are doing business, but it seems to be the lesser of two evils. I hope neither advertisers nor paid services are using AI/ASI for their short-term business interests and turning the world into a dystopia.

The Hopeful Path Ahead

I hope we are muddling through the coming years until we have AI/ASI becoming dominant in helping people. I hope the individual consumer is in Big Tech’s focus and that politics/legislation is giving us control over our data at some point. I hope inventors, developers, and entrepreneurs are giving us the tools to regain our autonomy. Where will this lead? Hopefully to a more self-determined world in which we sooner or later don’t need to work for making a living. We may see a world in which people want to work (part- or full-time) for having meaning in their life. And if people choose to live their life for short-term gratification and entertainment? … So what? It’s their choice.

Finally, with more and more business activities are being disrupted by tech, automation, and AI — and hundreds of millions, potentially billions of people falling out of traditional employment situations, I hope that ASI could create a complex network of reciprocal relationships that help people not only survive but thrive without the income that they were receiving from employment. I would not state this if I don’t believe that this is feasible.

Third World countries will feel the force of mass unemployment first and these governments have not that kind of money, nor the economy to pay a basic income. There is only AI that could help people to work and collaborate together so that the impact of that situation is mitigated by a (cashless) bartering for goods and services. Every person, even if he/she has no employment or cash, has time and the ability to do something for someone else or for the benefit of the entire community.

Therefore, I hope energy will become so cheap (together with other necessities: food, shelter, clothing, communication/ internet, health, and education) that even people living without a regular cash income could afford enough because they are using AI, delivered via smartphones. Much of these necessities could be produced or delivered for almost free if we want that to happen. There are already a lot of tools decentralized available — AI could help people to get the skills to make many amazing things happen in their local community with whatever resources are available.

We know that people after a severe disaster or after a war (like people the years after the end of World War 2 in Europe) were able to collaborate in a way so that their standard of living was much better soon after these severe events. AI could become an indispensable tool in facilitating collaboration and in turning small neighborhoods into (mostly) self-reliant communities, which may even exchange goods and services to each other in a cashless, but tracked manner – so at some point, there is a transaction and (delayed) reciprocity — instead of instant reciprocity as it is known from money. And then, few people with cash income could join the AI-supported bartering and help these communities even more.

Realistically, if it is up to the advertisers, then people without income are a useless audience – not worth any attention or help. I would disagree. In the long-term, once AI has stabilized their living conditions, more and more self-reliant communities could join again the consumer society and spend on goods and services maybe not immediately as individuals, but as communities.

In Max Tegmark’s book, “Life 3.0” (2017), he’s telling the tale of a company, Omega, an entity under the control of an AI/ASI that is making the world a much better place. I thought to myself, we don’t need to wait for the rise of an ASI to get similar results. Technology is powerful and transformative. We could start before we have ASI, and who knows, we might be along the way to a better world before ASI arises. The problem is, we don’t believe Big Tech or even Technologists are benevolent enough to create the software foundation to make the life of billions of currently poor people better — why helping them to optimize their life? Their time is so cheap – there is seemingly no value in that. I disagree and I explained it above. I am an optimist in believing that Big Tech and technologists will (at some point) deliver value to where it is most needed … and I hope that Big Tech becomes aware of the toxicity of their devil’s pact with the advertisers.