Humayun Sheikh is an innovation entrepreneur, a founding investor in DeepMind and is the CEO of Fetch.ai, an innovative platform for connecting IoT devices and algorithms to enable collective learning. Built on a high throughput sharded ledger, the Fetch.ai architecture delivers a unique smart contract capability to deploy ML/AI solutions for decentralized problem solving.
How did you initially become interested in artificial intelligence?
My background is computer engineering but I have spent my last 20 years in trading commodities, developing ML/AI algorithms for market trading, predicting markets, price prediction, and working with various AI projects to deploy commercially valuable usage to AI. I was introduced to Demis Hasabis of DeepMind 15 years ago and I got involved more with Artificial Intelligence in gaming. After DeepMind, together with my co-founders Toby Simpson and Thomas Hain, we founded Fetch.ai and our approach since then has been to build something that we could start commercializing.
You were an early investor in DeepMind before it was later acquired by Google, how did you come across this opportunity?
I was introduced to Demis Hasabis of DeepMind and we worked together for 3 to 4 years. During my involvement with DeepMind, we were pursuing ideas and ways on how we could make machines behave and interact more like humans. Ultimately DeepMind was sold to Google and is now one of the world’s leading organizations in Artificial General Intelligence.
DeepMind was famously attempting to create an Artificial General Intelligence (AGI), do you believe that building an AGI is still a possibility?
It may seem like a far-fetched possibility, but we also never imagined 20 to 30 years ago that AI labs like DeepMind would come up or companies like Google, Apple, Microsoft, Amazon, and Facebook would be investing so much into AI research, including neuroscience research. Capability-wise, we are still quite far away from achieving artificial general intelligence and then there is the question of commercialization. If Google had not stepped in, companies like DeepMind would have probably failed. Given the rapid rate at which AI is developing, we could expect reaching an inflection point when the AI research community surprises us with the development of artificial general intelligence.
Your latest venture is Fetch.ai, could you share the genesis story behind this startup?
I, along with my fellow co-founders, have discussed the ideas behind Fetch.ai for many years before we found the technology combination that would deliver it. We believe in a more efficient, optimized and decentralized world where identity, value and utility are returned to the individual and where collective knowledge is available to all users of the network. It turns the world into building blocks of exciting new businesses. The world is complex and underutilized: we simplify it and make better use of what we have.
One of the first Fetch.ai projects is a blockchain-based AI smart-city infrastructure project in Munich. Could you explain what this is and how blockchain and AI can assist a driver with locating a parking spot?
The Datarella-Fetch.ai Smart City field trials uses Fetch.ai’s AEAs (Autonomous Economic Agents) to unlock data and provide smart mobility solutions in Munich’s commercial real estate properties, for this field trial we have chosen the Connex building complex in Munich.
Each registered user who is a regular car park user is incentivized to reduce their individual traffic to the connex offices and for that he/she will be rewarded with a certain amount of tokens per minute for not parking at the parking lot. Autonomous Agents will autonomously negotiate the ‘price’ of parking spaces between the holders of them, and those looking for a space. As soon as a car or its related wallet address is registered as parked by the Carpark AEA, the token airdrops to this wallet stop. The number of tokens rewarded per wallet and minute depends on the current utilization of the parking lot.
The aim is to support the sustainable and efficient use of city infrastructure in Munich.
Another project in the pipeline is the AI Autonomous Travel Agents program. Could you explain what this is?
Fetch.ai’s Autonomous AI Travel Agents offers a decentralized, multi-agent-based system that provides personalized, privacy-focused travel solutions. Using smart contracts and AEA (Autonomous Economic Agent) which perform the booking of hotel rooms through a direct provider-to-consumer model, we want to ensure cost savings for both hotels and consumers by up to 10%. Furthermore, our application is integrated with Amadeus that offers instant access to over 770K hotels.
By bootstrapping to Amadeus’s servers, we can fetch a hotel’s name and location and other information needed to complete a traveler’s booking without using an online travel agency or metasearch site.
A hotel owner can launch our application without any set up costs and the approach is based more on pay-as-you-go. The hotelier’s software will interact with the software that consumers will use to shop. More importantly, hoteliers won’t try to sell all their inventory on this new channel powered by artificial intelligence. It will be an extra channel in addition to the current mix we see in the hotel industry.
The Autonomous AI Travel Agents application designed by Fetch.ai is not intended to replace existing systems in their entirety, but more to complement them. It operates safely, non-destructively, and in parallel to existing relationships that hotels might have. It delivers an alternative method by which bookings can be taken: one where the customer and hotel deal with each other directly, and one where a more personalized, better value experience can be delivered.
Will you be working with OTAs (online travel agents), if not how will you on-board different hotels?
No we won’t be working with OTAs. We will enable hoteliers to register on a software meant for hoteliers and access the Fetch.ai network from the second quarter of 2021 when we launch the application suite.
What types of machine learning algorithms are being used in these applications?
We can’t go into much detail here, other than the ML that gets indirectly used with other Fetch technology related to search and discovery. However, natural language interfaces, and continuous learning of what’s appropriate for any given agent collection (which involves, amongst other things, reinforcement learning and RNNs) is a big part of it. But there will be more!
Is there anything else that you would like to share about Fetch.ai?
We want to continue our focus on building applications, product suites with value for our community, our partners, and we continue to focus on that in a publicly visible way so that our community can share our successes. This builds value, utility and the growth of our community. Plus, users can get involved by investing in the FET token, a backbone of the Fetch ecosystem. It’s required to find, create, deploy and train autonomous economic agents and is essential for smart contracts, oracles, and the transactions needed to deliver the new digital economy.
And they can always interact with us across our social channels:
Thank you for the great interview, readers who wish to learn more can visit the above social media or the Fetch.ai website.
The post Humayun Sheikh, CEO of Fetch.ai – Interview Series appeared first on Unite.AI.