SingularityNET: in-depth interview with the crypto-related AI project

The Cryptonomist interviewed SingularityNet COO, Janet Adams to speak about the crypto-related AI project, their future steps for the AGIX token, the marketplace and much more.

SingularityNET interviews: what potential do you think your AI marketplace could realistically have? (at the moment it seems not so busy)

We created our marketplace because we saw some significant disadvantages for everyday developers within the AI space. Our decentralized marketplace has the potential to democratize access to AI, as well as eliminate traditional roadblocks which slow down progress within the field. 

True innovation rarely begins in big tech companies, but instead with visionaries, academics and students. However, the way the tech space is currently set up, advancements are reserved for larger corporations who have the resources to buy up or mimic ongoing projects, quite often turning these into profit-first projects. 

By creating a democratic marketplace with no central controller, participants can rapidly innovate, build and see the benefits without having to create a startup or sell their solutions to a big company. Our marketplace is creating an all inclusive AI space. 

How is it going with the bridge between Cardano and Ethereum? Some numbers and stats?

The Bridge development was started in early 2022 by our Bangalore team and later has been fully transferred to NAINT. The main idea of the product is to provide seamless and lossless one-type tokens transfer from one Crypto Network to another, e.g. between Ethereum and Cardano.

The first pair of tokens that Bridge started to serve were AGIX and NTX. We finetuned the algorithm, polished the UI, and optimized the API negotiation on these tokens. Later, NAINT integrated two more tokens: RJV and CGV. Later, HypC will be added as well.

Now we are working hard on connecting to the popular Binance Smart Chain network and «throwing» the Bridge between Binance and Ethereum starting with CGV token. Quite a hard and complex task, but we are sure that soon, we will be able to present it to the world.

The total amount of transactions from 2022 (since April) to 2023 increased by 68%.

Total volume of transfers from 2022 to 2023 increased by 50%

Can you tell me more about Desdemona? What’s the difference with Sofia?

Desdemona is a robot diva and the lead singer of the Jam Galaxy Band. She has played at a number of shows around the world, including at the National Association of Music Merchants (NAMM).  Herself and Sophia are quite different, with Desdemona having more of a sassy personality and alternative dress sense – fitting in well with her diva lifestyle. 

How is the collaboration with Ben Ditto going? How did it start?

The collaboration with Ben Ditto is off to an impressive start. Recently, in July, we successfully conducted the soft launch of Yaya Labs, a new spin-off project of SingularityNET, at WebX Tokyo. Yaya Labs serves as a creative studio, focusing on Web3 Projects and robotics. Our inaugural project involves shaping the personality of Desi the Robot, with the ultimate objective of propelling her onto the global stage as a prominent AI celebrity in the realm of pop culture.

Last year, Edwin Eyre, our partnerships lead at SingularityNET, contacted Ben Ditto due to his exceptional execution of concepts that blend the uncanny and utopian, encompassing elements of beauty and technology. Ben is a seasoned director in various mediums, including moving image, CGI, creative coding projects, augmented reality, VR and live experiences. He now holds the role of Creative Director at Yaya Labs. This emerging domain undoubtedly warrants close observation in the upcoming weeks and months. 

What do you think about Elon Musk and his AI projects?

Elon Musk has seen more of the future of AI further in advance than most tech executives. And some of his enterprises have been dancing around the edges of AI for some time – Tesla’s work toward full self-driving, NeuraLink’s work toward tools for probing human cognition. That said, the AI portions of Elon’s enterprises have not so far been the most successful things he’s done – his big proven successes have involved hardware design and innovation. 

I’ve seen no evidence that Elon or any of the folks involved in his enterprises have any big unique ideas with the capability to leapfrog toward AGI.  And I would say the experience with Twitter (er, X) should serve as a wake-up call to anyone who still had the idea that anything Elon touches will immediately magically turn to gold.   

He is a brilliant person willing to take risks, and one implication of this is that a lot of his gambles will fail, especially when they’re far from his areas of core competency. 

From what I have seen, the general approach to AI being taken in Elon’s world is basically drawn directly from the deep reinforcement learning “orthodoxy”, i.e. it’s similar in spirit to what’s being done in Deepmind, OpenAI and other similar places. I don’t see it as terribly likely that Elon is going to outpace these other organizations pursuing similar conceptual approaches.

As a separate point, as an AI researcher I happen not to believe that the deep neural net approach being taken by the Big Tech companies these days, and which Tesla seems very likely to attempt to emulate, is going to be successful at achieving the more ambitious goals of the AI field, such as Artificial General intelligence.  

My own research intuition is that, while LLMs and other deep neural networks will play a major role in achieving AGI, an AGI system with capability at the human level or beyond will need to incorporate significant additional aspects as well. This is what I’m working on in the OpenCog Hyperon project and the enterprise-AGI company TrueAGI, working closely with SingularityNET Foundation’s AI team, in which deep neural nets are being hybridized with logical reasoning and evolutionary learning engines within the context of a large decentralized knowledge graph.  

The AGI research community includes a variety of other non-conformist approaches to creating advanced AI, however Musk so far is not going down this sort of innovative road, instead the indications are that he’s going to apply a mainstream Big Tech deep neural net approach but leveraging data to which he has unique access, such as data from Tesla cars or Twitter.

What are your forecasts (5 or 10 years) for the AI ​​services market?

I believe that within 5-10 years a breakthrough to human-level AGI will be made, which will cause massive disruptions to global society and economy, far beyond what we’ll see from ChatGPT or other generative AI systems. After this breakthrough, the AI services market will become the largest portion of the global economy.  

An interesting question, however, is the dynamics according to which this will unfold during the perhaps small number of years in which AGI advances to and then beyond the human level. Which industries in which regions of the world will AI services first pervade in a huge way, and which will lag behind?   

This cannot be known for certain given the manifold uncertainties in both the precise progress of technology and the global economy itself, but it seems clear that those industries that are already more thoroughly computerized and internet-ized will be dominated by AI services sooner. So for instance, finance and medicine and supply-chain will happen faster; whereas education and agriculture will be slower.

What are the incentives provided to developers to sell their AI tools within the SingularityNET marketplace?

Our marketplace creates an opportunity for AI developers to own and be fairly paid for their innovations. AI development is resource heavy and can often only be undertaken at scale by large organizations. However, our marketplace creates a new avenue for developers to be paid fairly for their work, instead of having to either seek employment from large corporations or sell their ideas and passion projects on. 

What are the growth prospects for using frameworks like OpenCog?

At present, artificial intelligence systems have narrow functions – meaning they can complete one task very well but cannot do anything else. The aim of OpenCog Hyperon is to integrate many different types of narrow AI, creating a system that is capable of many different tasks. This will allow us to achieve artificial general intelligence (AGI), technology that is capable of completing any task above the level of humans. 

What are the next big events that will be proposed on SophiaVerse?

There are many exciting events coming up for SophiaVerse, including the public release of Sophia’s AI Lab and a gradual roll out of features. We also have a host of new team members on the way and there is more information about the GUARD token to come.   

Do you think that the future of AI for singularityNet will mainly focus on software or will hardware devices also be integrated? 

The core of what we need to progress from narrow AI to AGI, AI that is capable of completing tasks above the level of humans, is of course the right cognitive architectures and learning and reasoning algorithms. However, without the right hardware, even the best mathematics and software can’t run efficiently enough to have practical impact.

AI development is expensive at present due to hardware constraints, meaning the space is largely occupied by larger corporations. For example, the current advancements being made in the field are largely down to the use of graphics processing units (GPU), which help devices handle graphics, effects and videos. Further hardware innovation is required for more breakthroughs to happen within the space.

This is what inspired us to partner with leading hardware developer Simuli. The partnership will focus on the creation of a Metagraph Pattern Matching Chip (MPMC), a new piece of hardware that embodies the two key knowledge-graph search algorithms, Breadth-First Search (BFS) and Depth-First Search (DFS). BFS and DFS algorithms are important building blocks for AI systems, as they are used for knowledge representation, reasoning and decision making. Combining these two algorithms into one chip will allow for better AI that is capable of completing tasks to a higher level than is currently possible.

What is the main use case of the AGIX token and what are the forecasts for the future?

Our AGIX token is the key to unlocking endless possibilities with artificial intelligence. Using it, AI curation, self-organization, problem solving, and transactions are made possible. Our platform and token serve the future of AI, allowing anyone to participate in and monetize AI, meaning companies, organizations, and developers can buy and sell AI algorithms at scale.

What’s your opinion about WorldCoin developed by Sam Altman?

I thought Vitalik’s in-depth commentary on WorldCoin made a lot of sense.   His comments seem to have been interpreted as negative by much of the blockchain media, but actually I thought they were mostly positive and very well thought out. 

A key point he made was that there can be no single silver bullet solution to identity verification and management – one will need multiple aspects to deal with the complexity of trust in the current social and technological landscape, and WorldCoin tech could possibly contribute to the mix considerably. 

It would be a mistake to put total trust in managing or verifying one’s identity in any one company, project or technology, but WorldCoin may well earn a major position in the global web of trust in this regard, that remains to be seen. 

It is a real and innovative technology with significant potential to do good for the world – if it is appropriately integrated into a context of a decentralized identity ecosystem including a palette of diverse tools rated and coordinated by a collection of decentralized reputation systems.