While the above ideas all sound super interesting, I think some SNet implementations of basic benchmark algorithms would be useful (Resnet, for example). This would allow people who are interested but skeptical to directly compare performance of a model on SNet to a standard implementation, which could speed up adoption in the long run
2 Likes
The second thing, which would be more ambitious but I think important for moving toward AGI.
You have acknowledged in the past that the singularity net at the moment has a bunch of narrow AI services. I would like to see more general algorithms and the agents ranking one another / utilizing each other’s services be fully fleshed out and developed. Working on the things you would need to have in place such that someone can send an email with or without attached relevant documents and receive advice/guidance/solutions from singularity net.
1 Like
Hey!
I’m a noob. But, I saw that people were thinking about how AGI can assist with an interface for jamming.
One thing that is constantly a problem in jamming is mismatched volumes / levels on instruments.
Can we get a simple algorithm going that automatically balances people’s levels while they play? Kind of like your own AI sound-guy for a virtual jam space?
Jason.
Third idea.
Create a superior search engine that uses singularity net. Something that is not driven by what will sell and instead uses your data to interpret what you are actually trying to find. Then uses all the other popular search engines perhaps, and returns a subset of relevant links.
To explain a bit more what I mean by this…
Unless I specifically search for academic papers, or I search for a paper by name, I am probably going to receive a bunch of media spam / sensationalism if I look up " developments in AI ". But if anyone where to look at my search patterns, either in google or in youtube and see all the calculus and linear algebra videos and content I have been looking at, it would be very evident those clickbait articles are not what I want. I am suggesting we could develop something better, something that gives you what you want not what they are trying to sell.
3 Likes
Hey!
How about a “bandwidth advisor” or “bandwidth monitor” for small companies.
A simple AI solution could be a company bandwidth monitor that tracks your company’s internet usage and shows a “green light” for sending things to the printer or “red light” to avoid sending things to the printer. It could track real-time indicators around printing, downloading, emailing etc. to help day-to-day small business operations run smoother and make suggestions going forward.
Huge problem at our workplace.
You might say that we just need to upgrade the internet but the reality is that’s a huge cost for small companies who are struggling to pay salaries. I think we could benefit from a simple “Bandwidth advisor” at the workplace, or anything that helps us be more efficient with our technology.
How about a “tech advisor” or “tech sales advisor” for companies.
One step further for AI would be an “advisor” that tracks sales and shipping costs online for upgrading your company technology. So, if the “advisor” thinks you need to update a certain computer’s RAM (it could automatically detect) and finds a sale on some tech (you give it your budget), it lets you know that your company could benefit from a sale and asks if you would like to place an order.
That would be remarkably helpful to many, many small businesses.
How about a “proof-reader” service that works with any word processing software and helps your company to correct mismatched fonts, adjust margins, adjusts mistakes in numbering, adjusts number spacing, tabbing, spelling, etc., that works with your own company’s style.
A huge problem in our small company is that we are constantly spotting errors after the fact that are way beyond current word processing capacity. Someone once said that word programmers must be actively “trolling” us. The word processing document does not question a use of font, spacing, margins, or certain spelling, etc.
Whatever templates you do use are easily able to be over-ridden while an employee is inputting their data, etc. and some of our older docs are completely mismatched to our current ones. In small companies even our headers are mismatched.
An AI solution that could “match or synchronize” all your docs, reference pages, footnotes, etc. for formatting that learns from your own company docs without having to endlessly train a template or harass your staff about how to use word processing programs could be very helpful for almost any small business, period.
Jason.
2 Likes
Hey!
One other solution that I think AGI is already working on would be to merge the dictation (voice-to-text) software with read aloud data so the read aloud begins sounding like a more natural speaker. Even better if it could read it in your own accent back to you (creepy right!).
So, when you ask the .pdf to read aloud it tracks native speech inflections in the language for you and reads it back to you in a way that you’re used to hearing it.
Put more succinctly, can you give Sophia a Brooklyn accent by combining speech-to-text with read aloud?
I think that’s a fairly realistic synthesis to ongoing projects.
Jason.
I can visualize how this has the potential to make a huge impact on everything on this planet and change life as we knew it. I absolutely welcome it and I feel incredibly grateful to have even found this.
I have zero programming experience but I have nothing but time and a thirsty brain anymore. My days of “honest work” are all done. As far as scaling down into all the micro transactions and not “jamming” the network…well can Celer tokens be implemented into that aspect? Also…there’s a guy I’d love to drink a bottle of bourbon with that is cluttering the sky with high speed global internet satellites. That is an infrastructure that’s made for a project like this…Especially if our future is going to have robot doctors doing house calls. Those machines will need to stay connected in all conditions. Sorry for all the run on sentences and poor punctuation guys. I too poor to actually stake enough to have a vote in anything since my injury but I’m staking literally every spare penny I have. Hopefully I can help contribute to this project in some way though. I really want to be a part of a positive impact. I believe this can absolutely make the world a better place for everyone if enough positive decisions are made. This has way more potential than just profits. Money doesn’t make people good people. We’ve all seen what greedy monsters with power and money do. Not that all people who have money and power are monsters. I just really love that this is going to flatten the drive for war for profit. Or it has the potential to anyway. Thanks for coming to my Ted talk.
Would it be possible to make a partnership or work directly with neuralink? Perhaps have the devices act in part as computation devices and partly as receivers / senders / users of singularity net AI? None of the existing algorithms seem appropriate for that use right now, but presumably, neuralink could be interested in making use of the emergent higher level problem solving we hope to achieve with singularity net. Given that’s what we are aiming at, does it not make sense to design the systems from the start to be compatible? Does it not make sense therefore for the algorithms they need to exist within the ecosystem so that they can be bootstrapped by the other algorithms as they develop?
Browser plugin for reliability-detection of content.
Powered by a reputation engine for public content
Arguably, one of the largest short term threats to society are echo chambers and information bubbles that create their own reality that is not based on facts. It’s troubling to see how some of these communities drift away from mainstream to become extreme and violent. either in their communication or in a physical form. Therefore any initiative that can counter this phenomenon is worth funding. At the same time we have to protect 'healthy alternative ideas and enable them to prosper.
I don’t have anything near a final solution, but i’m pondering about a solutions around these lines:
- Create a browser plugin that tries to relate the content of an article to other articles of the same topic, that have earned a high reliability score
- Perhaps a sufficient capable AI can list the main arguments and counter arguments on the topic of the loaded article, and visualize the reliability of each argument.
- Ideally the plugin can measure the loaded article itself for trustworthiness. (A measurement that can change over time of course)
The plugin (or whatever form it will take) should not annotate an article for the kind of views it is giving (right, left, extreme, etc), since that will only motivate people to stay away from counter views.
Problems to solve:
- how to measure reliability in an open environment e.g:
- By starting with some diversified collection of ‘reliable’ sources
- By collecting peoples reading habits and give a higher score to those that have a more varied browsing / reading habit,
- How to motivate people to install the plugin
- Perhaps start by trying to collaborate with Brave)
- how to motivate people to use the plugin. (since it takes more cognitive load to study alternative views than having your own views enforced.)
Note: this plugin will not be able to change the views of already radicalised people, since they will see it just as another conspiracy, but it may lead people away from that road before the become radicalised, or help them to move away, if they are already in doubt.
1 Like
It would be really cool actually if the AI basically highlighted the article for you, marking red the things it finds questionable or unsupported, and green the points the article makes that it can verify.
Charles from Cardano talks a lot about these sorts of things. He promotes some kind of certification people can put on things they link to verify that they have done the research and can vouch for the legitimacy of the content. His ideas go beyond this, but this was a particularly good solution I thought for social media related communication. < Tie people’s reputation tightly with what is being said such that people hopefully don’t spread as much non-sense.
I think there is room for the AI to assist in this.
I would like to see SNet move away from selling things to people personally. Yes provide things people want for a token cost. But I would not like to see this doing even more of what is already happening >> promoting consumerism by targeting people patterns it can find in data.
Hey Brett,
Thanks for your reaction! That is a very valid remark!! I agree that promoting consumerism is not at all desirable. still, I do see 3 benefits:
- Making the current mechanisms les ‘bad’ and more people-focused. Imagine the tool recommending sustainable alternatives for regular products. Actually: this could even be a better, more narrow application in itself, hence marked in bold.
- Simply getting traction for SNET by turning current mainstream applications decentralized.
- Giving smaller retailers a fighting chance against the Amazons and Alibabas of this world.
I guess there are many details that determine whether such a tool will be truly beneficial or indeed just a slightly less bad or indeed even worse alternative for the current tools. Would be very interesting to brainstorm on how we can actually make this into a good thing. this could be a phased approach where we start closer to the current offerings, and move it in a more beneficial direction step by step.
I agree with you that we want more mainstream applications to become decentralized. Making people’s data anonymous, but still available for analysis, I agree although it would be nice to at least be able to specify for what purpose my data could be used. My issue is really just with the sales side. I would think there are other applications where those two ideas can be applied. Medical often pops up when I have heard others’ conversations at this point, and I think probably rightly so.
I really like the anime character idea 
sounds fun.
Streaming voice to text also sounds like a great and narrow AI project.
@Brett_Whiting I agree. What Ben is talking a lot about, also together with Charles is to construct a kind of reputation machine that will filter out noise from Social media, such as unfounded claims or outright lies and insults. The first challenge however, is to get the existing social media networks embrace this, or create a new one that should then first get enough traction.
So basically this idea is a shortcut by trying to add a similar process on public data (including social network messages, if possible/crawlable). Of course that makes the whole process much harder and probably less reliable, so it is not a substitution. But who knows, what kind of clever things we may come up with!
Hi Juan,
This idea is brilliant and, in fact, I have had similar ideas in the past too. I even work for a small start-up in Thailand that has this as a main goal. That is to provide a platform that allows farmers to form a non-zero sum game where the data they provide helps create the next best deep learning model to identify diseases on leaves, pests or even just the general health of the crops.
My startup is moving towards robotics and we want to eventually build guardians that scout the farms and tend to the problems. Combined with satellite and drone data this is a revolution that could help feed the world. Decentralising this on singularityNet is the most utopian way of achieving these goals!!
I would be happy to talk to you about this.
Ethan Connelly.
2 Likes
I’m glad to hear that there are people already working in this kind of things.
With my limited knowledge, I think Singularity net would be a good platform to integrate different services that belong to the agricultural ecosystem and make them interact with each other, right like in nature everything is part of a complex system and each part help balance the others.
For instance different services (even from different start ups or developers) like disease diganose and drone surveillance can be combined with another water management app and even analyze data on how temperature is affecting the whole.
I would love to discuss this with you and with whoever in the community is interested so I started a new topic in the projects section. See you there!
2 Likes
DOTA on SNet is a cool idea…
In a similar vein, in the OpenCog Hyperon proto-AGI project we are experimenting w. Hyperon for carrying out various missions in Minecraft. Part of the goal there is collective intelligence among multiple gameplayers. It would be cool to make multiple Minecraft gameplayers wrapped in separate SNet agents, playing the same game cooperatively/competitively…
I never played DOTA, I wonder if it supports multiple cooperating / competing players in complex alliances?
A friend advocated Advanced Tactics Gold as a testing-ground for early-stage AGIs but I never played that game either, I have been deficient in game-playing since my first batch of kids grew up…
This makes a lot of sense but it would be ideal to break it up into a bunch of sub-suggestions. Actually OpenCog would be an ideal framework for merging together data from different agriculture services. But making a nice API-based deployment of each individual agriculture service is a decent-sized project in itself I think. (As often, the real work is not in putting the service on Snet, the work is in making a service that’s cleanly API-based so it can get wrapped in Snet) … Do you have any specific ideas about what software would be used for the individual agricultural services you’re thinking of?
SingularityNET Foundation and Arizona State University and Leshan Agricultural institute (in Szechuan) have been talking about similar stuff for a while, e.g. related to automatic diagnosis of crop disease, and estimation of degree of natural carbon sequestration in farms and forests etc. But progress has not been super fast so far, it would be great to see stuff accelerate in this area.
1 Like
@Ethan_Connelly this sounds very promising, perhaps you can contact Dr. Matt Ikle’ (matt@singularitynet.io) who coordinates SNet Foundation’s agricultural-AI projects to discuss…
2 Likes