How SingularityNet Leads to “AI for All”

tl;dr: The democratization of AI has already begun. A decentralized AI marketplace may supercharge innovation and truly level the playing field.

I’m giving a presentation at the end of March at an event for some of Microsoft’s biggest customers.

If you have watched any live sports in America recently, the topic won’t surprise you. AI, artificial intelligence. They are all-in.

So, in my efforts to avoid total humiliation on stage, I’ve been on a multi-month journey to understand, or rather, begin to understand AI.

It’s made me feel a bit sheepish honestly.

I’ve long maintained that blockchain is the entree after the appetizer that was the Internet.

The problem with that analogy is I’m not sure how to describe AI.

It may be the restaurant in which both of those are served.

It’s that big.

I mean, there are already AI-generated modeling agencies . They are not quite ready for prime time, but when Generative Adversarial Networks (GANs) can create people that look real, but are not, and “deep fakes” can create compelling video, you know that society is about to change pretty dramatically.

Of course, what could be bigger than either blockchain or AI?

Well, blockchain + AI, of course!

And that’s why I’ve been paying particularly close attention to a project called SingularityNet.

The Great AI Democratization

To understand why I am intrigued by this project, in particular, we need to take a step back a bit to provide some context.

One of the trends I’ve discovered in my research for the Microsoft presentation is the The Democratization of Data Science.

I touched on this a few months ago in the post Why You’re a Data Scientist…But You Just May Not Know It Yet. However, the proliferation of tools like Keras, FloydHub, and even Microsoft’s own Cognitive Services is making AI increasingly accessible.

You still need a fair amount of technical skills to make these things work, but the trend is clear. More AI power will be available to more people at lower costs than ever before.

This will create tremendous opportunity, but results will be all over the place.

As Trent McConaghy of Ocean Protocol said to me,

“there will be equality of opportunity, not necessarily equality of outcome. There’s still a place for hard work.”

The democratization of AI will also necessitate an up-leveling of skills, which the HBR article touches on. That’s pretty important for the future of work.

As the article says:

Very few companies expect only professional writers to know how to write. So why ask only professional data scientists to understand and analyze data, at least at a basic level?

HBR

The AI Competitive Landscape

More important than the future of work though is the future of something else.

The world.

You see, one of the challenges with AI as it currently exists is that the algorithms behind them are “closed.” That means you (or anyone) can’t inspect them to understand how they work or check for bias. This is a HUGE deal in AI.

When AI is “closed,” it means that each of us is vulnerable to manipulation by forces beyond our control and over which we have no influence.

It doesn’t end there.

What’s more, when AI remains the sole province of large companies (yes, like Microsoft), three things result.

The first is that “the rich get richer.” Since data scientists are in short supply, the demand is through the roof for them, which leads to 7 figure salaries . That’s not exactly something that small or mid-size companies can necessarily afford.

The second is that the cost of AI services, despite its democratization presently, becomes costly. After all, the large companies have to pay for those big salaries somehow.

The third is that innovation is stifled. With algorithms being proprietary, there are limited opportunities for the combinatorial innovation that comes when two ideas are put together. Or, worse, people who can be helped are not helped, as we discussed in this post about Cigna’s algorithm that predicts opioid addiction.

So, while there is definitely room for players like Microsoft to do a lot of good with their AI solutions as they already are and which you can see here, there is also a need (and dare I say a requirement) for something more.

SingularityNet and the AI Marketplace

This is where SingularityNet comes in to the picture. With near perfect timing, they raised $36 million in 60 seconds in their ICO  back in December, 2017.

The goal is to build a decentralized “AI-as-a-service” marketplace that allows anyone to very cost-effectively requisition AI capabilities for their business.

What’s more, and perhaps of greater significance, is that they give AI programmers a way to directly monetize the value they create.

Since it is a decentralized marketplace, there is no middleman. That reduces risk and lowers fees for all involved. Plus, because it is open, combinatorial innovation is unbounded and verifiable.

An AI marketplace like this has the opportunity to both democratize access to advanced capabilities while not becoming the domain of a handful of people.

Given how important AI is going to be in the future (and already today), this type of security for humanity is critical.

To understand why, I highly recommend this excellent talk by SingularityNet’s CMO, Arif Khan. The way he provides a historical context for the choices we are all going to encounter in the future is extremely compelling.

You Can Use SingularityNet Today

One of the things that I love about the world of crypto/blockchain is that the products available today not only did not exist 2 years ago, they could not exist two years ago.

Yet, here they are.

So, today, if you want to, are fairly dedicated, and have patience, you can read the very clear Getting started with SingularityNet guide that Arif put together. Be forewarned, it’s extremely detailed, but there are a lot of steps here. Of course, that’s the price you pay for living in the future.

When all is said and done, you will have access to the beta version of the AI marketplace which is live on the MainNet of Ethereum.

And you will have a foot in the future of AI at the same time.

SingularityNet Beta marketplace

What I like about the marketplace is that the steps required to consume services are explained in clear English, a welcome change from many other beta sites.

I paid for an image recognition AI service that identifies dogs and flowers. As you can see there are others and more will come, particularly as AI developers begin to revenue potential and the demand side sees the low cost, democratized access option.

As you can see, there is a “thumbs up/down” voting system. For now, anyone can vote on anything, but in the future I suspect it will be more closely aligned with paid usage.

In the long-run, this could become a vibrant marketplace where the top quality AI services are available for consumption and integration (paid in AGI tokens), vetted by a verified community of users, and open for all to inspect.

Also, I would be remiss if I didn’t mention, Dr. Ben Goertzel, the CEO of SingularityNet, who has as much street cred in the AI world as anyone on the planet.

He’s the real deal, with a balance of vision and pragmatic execution, as this note from the community Telegram group demonstrates:

“The beta doesn’t come with a good example of multiple AI agents working together in a way that yields interesting/significant emergent “society of mind” type intelligence…

It’s a bit simpler than that to be honest, but it is a solid decentralized platform that AI services can be integrated with … 

and it does provide tools that make it straightforward to connect multiple AIs into a society-of-minds (well, straightforward if you use the python APIs … if you use other languages then in the beta version it’s still complicated, but we will beef up the support for other languages a little later)…. 

We will be putting more complex AIs out there on the network as the year progresses, the beta is just a start. 

But it’s a new phase of development in that it’s a platform that is really solid and scalable to build on — including building emergent intelligences as well as all sort of commercial AI services with simpler architectures, etc. 

We have a lot more cool/complex AI in our research codebase but what has been put into the beta initially is just a fraction chosen mostly for ease of use and ease of integration not underlying sophistication….

For sure we are working toward the same vision as articulated at the start of the project, but when really building stuff rather than only talking about it one’s gotta go step by step…”

The AI Future Isn’t Coming…

It’s Already Here

One of the quotes that I put into my upcoming Microsoft presentation is from the Atlantic Council’s AI Research Group:

There is a first-mover advantage of AI, as the global leader can gain an economic and military edge over rivals and will be in a position to write the rules that govern AI’s uses.

While I don’t think there is such a thing as “writing the rules” for AI, since we can’t predict or really control how AI will evolve, I do think that first-mover advantage is very real.

When a marketplace like SingularityNet comes online, the companies and individuals that figure out how to leverage it quickly and effectively will generate substantial first-mover advantages.

That alone could be the difference between “winning” and “losing” in a given market, eco-system, or even geo-politically.

It’s no wonder SingularityNet had 10x the amount of interest for its token as it was willing to offer.

I encourage you to keep an eye on it.

Here are some videos of community members who have used the beta

P.S. There is a fantastic SingularityNet podcast that is one of the best I have heard. Very much recommended.