AI for AI – feels a bit like inception, but artificial intelligence is now being used to optimize AI models. Now, this is not self-aware AI, but it is aware when new AI models are introduced into the ecosystem. So, what are the benefits?
First, these new AI solutions are reviewing the AI code base to streamline it and make it more efficient. Second, this has shown a downstream impact on resource consumption which creates cost savings for organizations. This also feeds into the global sustainability efforts to reduce the overall carbon footprint by using resources more efficiently.
Further, we are seeing a convergence of AI and blockchain which will create explosive impacts on financial services and crypto.
Joining me on the podcast to dissect all of this is Dr. Leslie Kathan, CEO of TurinTech and Researcher for UCL Centre for Blockchain Technologies. We’ll explore this shift in AI, how blockchain is disrupting traditional financial institutions, and how our emotions factor into financial decisions.
01:31 – Dr. Leslie Kanthan has an extensive background in machine learning, artificial intelligence, and graph theory. Additionally, he has over 10 years of experience in financial services and automated markets, making quantitative trading, quantitative research, and data science.
02:22 – TurinTech was developed based on a paper that Dr. Kanthan had published with three co-founders called, “Darwinian Data Structure,” which scientifically proves that they can optimize code.
03:44 – When optimizing AI, what is the expected outcome? By using cutting-edge techniques and models, the goal is to reduce the memory, the CPU, and the energy consumption of code.
06:50 – As new code and models are frequently introduced, AI for AI will compare new models to your own models. So, the AI model is being improved as it’s going through the learning process. AI for AI is like a symbiotic relationship; it requires optimization and continual learning.
08:07 – The processes with the AI models involve optimizing data structures as well as parameters. Because there is full transparency with the process, machine learning can also be applied to the strategy to continuously improve the parameters to continuously improve efficiency and accuracy.
09:44 – Considering explainable AI, the outputs show what was understood, what it’s doing, and what information it did ingest. Once there’s explainable AI, it’s easier for non-tech users to understand.
11:00 – As the blockchain is emerging, it is also causing disruptions to the institutionalized traditional mindsets around financial services. Artificial intelligence and distributive technologies, specifically blockchain and cryptocurrency, are the two frontiers right now. The survey paper regarding cryptocurrency trading combines both elements of AI optimization and how it can be applied to cryptocurrency trading using sentiment analysis.
13:39 – There’s much volatility that there are individuals and elements that can move the market and prices. Dr. Kanthan explains how they use their model generator and optimizer to effectively model this using NLP and using sentiment analysis.
16:23 – While many are still trying to figure out how to better leverage blockchain and financial services, the emergence of DAOs is causing a ripple effect and some uncertainties in the crypto world. Essentially, its purpose is to allow for approval and authentication through a group member voting system—enabling it to be regulator friendly and providing fairness.
20:17 – Sometimes it takes a force to get companies and people to shift and accept the digital-first perspective. There’s going to be a convergence of all these technologies coming together—cryptocurrency and blockchain, AI, and even the metaverse.
22:46 – Now, the metaverse is quite visionary—but there will be people who will grow up with it already having existed and be familiar with that technology space.
23:39 – The share price of the largest hardware suppliers of AI is going through the roof because the consumption for that hardware is massive. Consumers that are using this hardware have numerous use cases and thousands of models.
25:00 – There is profitability, but is AI costing a lot right now economically? If you can cut down the cost, reduce energy consumption, and maintain your targets, then you can achieve good profits and continue to reduce costs.
27:00 – Implementing AI is not intended to replace people—it’s to augment and encourage people to be more efficient through upskilling.