Cosmas Wong is a tech entrepreneur and investor who created companies for more than 20 years. He is the founder of Grey Jean Technologies; the genesis of the GNY platform, and also of other strong technology companies. I had the opportunity to make that interview with him, and we discussed the problems of AI, blockchain, and crypto adoption as well as his business ideas.
Andrey: Let’s meet with the Hacker Noon community. What is your background?
Cosmas: I’m a corporate lawyer by training. I spent over 13 years in mergers and acquisitions, private equity, corporate finance and venture capital investments before I started my first company in New York in 2010 with a few other co-founders.
Andrey: What companies were built under your management?
Cosmas: The company, Enso Financial, is a financial services data analytics company, and it was sold in 2016 to a financial services group in the UK. In 2015, I started Grey Jean Technologies in New York, and we started to build the Genie machine learning platform.
Andrey: Why machine learning?
Cosmas: The main catalyst for Grey Jean was to find a way to organize the large amounts of unstructured and semi-structured digital data that we produce every day and see if we can find repeatable patterns that might predict our next actions. This depended on having a process that can find correlations in the datasets that the human mind or eye cannot detect; correlations that don’t sit within an X:Y or X:Y:Z, but a multi-dimensional axis, and one that changes all the time. Only a machine can see and track that.
Andrey: What are the challenges of blockchain adoption?
Cosmas: Not enough people understand blockchain, how it’s different and better than what we currently have. Our opportunity is to show the implementation of blockchain technology more widely and provide more useful real-life use cases.
Andrey: What place blockchain will take in everyday life?
Many large institutions — banks, insurance companies, governments — are testing blockchain and cryptos. When we see the training wheels come off, adoption levels will skyrocket. When that happens, and comfort levels increase, blockchain technology will simply be another technology that operates in the background of everyone lives.
Andrey: What are the barriers to AI adoption that the tech community should overcome?
Cosmas: The biggest barrier to AI adoption is cost. Cost of implementation and cost in time to implement. All too often AI technologies like machine learning are seen as “cure-all” panaceas for every single data problem; companies think they can throw all the data they have at a solution and expect to get every question answered. We have seen that that is a sure way to failure. AI solutions are intelligent, but they are artificial, and they require training and they need to be tuned to answer more specific questions. They need to have the right data sets. That is our approach; define the question to be answered, test the right data sets, and expand from there.
Andrey: Elon Musk said that AI can be a fundamental risk to the existence of human civilization. What do you think about his warning?
Cosmas: I admire Elon Musk for everything he’s done, but I think that’s a complete overstatement. The late Stephen Hawkins had the same opinion.
We equate AI to some all-knowing all-doing mega computer brain like “Skynet” from Terminator. I’ll agree that “Skynet” is a fundamental risk to the existence of human civilization, but we’re far from it. Having said that it is true that AI solutions are a fundamental risk to certain types of jobs — jobs that are repetitive, certain jobs that require immense quantitative requirements, jobs that require better efficiency — but that’s also what they said about algorithmic trading platforms.
Although certain types of jobs will be eliminated by AI solutions, the human functions for those jobs will also change. The people who succeed are the ones that train themselves to master the machine, and let the machine do all the heavy lifting.
Andrey: There is a lot of buzz in the media about Facebook’s new currency Libra, but many consider it wrong use of blockchain because it’s not decentralized. What are your thoughts on Libra and the use of centralized blockchain?
Cosmas: I think the launch of Libra just catapulted the awareness of cryptos into the mainstream, and that’s a good thing. We can debate whether it’s a real crypto because it’s not decentralized. As I said earlier, blockchain technology needs to be more accepted, and the technology needs to morph and change and go where it’s needed, and we need to let that happen. If some want to use it in a way that’s more centralized, then it still serves that purpose.
Andrey: What is your main business and leadership target right now?
Cosmas: In September 2018, we launched GNY.IO with new investors and partners, in order to decentralize and distribute the machine learning platform on the blockchain; we called it GNY.
Andrey: And how is it going on?
Our first major milestone this year was the GitHub release of the core GNY code. The next target is scaling GNY’s capability on a chain with a release this month. Developing the scope and capability of GNY will be in constant evolution along with heading towards our sidechain solution for token deployment and onboarding partners and developers to the system.
Andrey: How did you come up with an idea to combine machine learning and blockchain?
Cosmas: The idea to combine both cutting edge technologies came out of necessity. We saw that some of the challenges in machine learning could be overcome by the blockchain.
A typical machine learning platform relies on being able to churn a huge amount of data, and all that data needs to live in one place. That’s why many large enterprises are building data lakes. With the increasing push towards data security and ownership, we believe the opposite will occur — that people’s appetite for their data to be shared and collected, especially personal data — will not be centralized, but distributed.
We’re also seeing that although machine learning solutions are becoming more popular, and more people are testing them, they are also less likely to want to give their data to a third party to be processed. Centralized data and transferring sensitive data to be processed, means a higher risk that a security breach of that database will compromise the security and privacy of more people.
By distributing the machine learning system on the blockchain, we are letting a company or a group or consortium of companies that normally share data, to build a GNY side chain so that the necessary data is shared and used securely within that group, and all processing is done on-chain. Nothing leaves the security of the chain.
Andrey: Machine learning system requires reliable and consistent data, how will you ensure this?
Data entering the blockchain will be verified and exist in a form that the machine can learn to read from the beginning.
Andrey: A typical machine learning platform requires millions in investment and a long integration period, can you solve that problem?
A distributed system like GNY can be deployed much more quickly and run more cheaply. This is extremely important for us as it promotes the use of the technology and “democratizes” it so it can be used by not just the very large corporations.
And by making GNY lightweight in order to sit in the blockchain nodes, we are amplifying its power by using the blockchain’s natural distributed infrastructure as a neural net.
Andrey: Did you test GNY somehow? Do you have use cases?
Cosmas: We’ve had the privilege of testing the predictive powers of our platform with some large retailers and publishers and have consistently showed very impressive results. For example, we compared our predictive platform’s ability to predict a publisher’s website users’ interest in certain articles versus their own editors, and found that we were consistently between 90% and 120% better. We could also predict a large US retailer’s customers’ next day purchases at over 70% accuracy. We will be releasing more detailed use cases later this year.
Andrey: Are there any unique technologies that your company uses in your neural network?
Cosmas: GNY’s recurrent neural network deals with sequence problems because our blockchain connections form a directed cycle. In other words, GNY can retain state from one iteration to the next by using our own output as input for the next step. In programming terms, this is like running a fixed program with certain inputs and some internal variables. The GNY recurrent neural network can be viewed as a fully connected neural network if we unroll the time axes. We’ve filed a provisional patent for this technology in the US.
Andrey: And the last question. We can see that cryptocurrency little by little fight their way to everyday life. What is your opinion, crypto community and ordinary people are ready now for a global adoption?
Cosmas: I actually think that blockchain technology will integrate into everyday life. Certain blockchain products and operations will soon rapidly demonstrate their superiority to the established best practice in technologies.
Similar to how the internet increased digital interconnectivity, blockchain will elevate digital ownership, activation, security, and optimization rapidly. This will force new standards in an already competitive digital marketplace.
Currently, we are onboarding our first beta-commercial partners and it is incredible to see how quickly they are moving from, “How can this technology improve my business?” to, “What else can GNY do to unlock value and meaning in my existing data operation?” We are most excited to see how small to mid-sized enterprise use GNY to build new and improved digital operations. We can’t wait to see what people come up with.
I’m also excited about the greater and more widespread adoption of cryptocurrencies in everyday life. A decentralized “store of value” is crucial in our current trade and political climate. Having said that, we also need a much more stable cryptocurrency, and bitcoin’s wild daily fluctuations don’t help. I believe as more people adopt cryptos, they’d stabilize, but again, we’re a long way off.