This Analyst Note charts the rapid pace of advancements in emerging technologies - AI/ML, Blockchain and IoT, across the world and assesses the resistance for change in the existing industries and companies and recommends change in their conventional playbooks.
While attending the MIT and Mint’s EmTech India 2018 conference earlier in March this year, I heard one of the speakers end his presentation claiming “This is the fastest pace of technologies development that we have ever witnessed in our lives… and, will be the slowest we will witness for the rest of our lives”. Now, for some of us who track the technology and market trends, the first half of that statement is hardly a surprise, but what was really intriguing was the second half. EmTech India brings eminent, leading global experts on Emerging Technologies (AI/ML, Blockchain, Robotics, IoT and Quantum Computing) together, and is one of the best tech conferences in India. And, this proclamation on such a platform opened up conversations among a number of delegates.
Emerging Technologies are Advancing at Rapid Pace
As we take note of all the advancements in Emerging Technologies over the past year, we witness a number of significant developments:
- At Convergence Catalyst, we published our Artificial Intelligence industry report exactly a year back in June, 2017. Then, the research for our report revealed that it takes anywhere between 19 to 40 million pictures of cats for an average unsupervised deep learning algorithm to recognize a picture of a cat, unaided. A follow up research on the same topic earlier in February this year pegged the number down to less than a million. This became possible thanks to the significant research and advancements in Low-Data Techniques, Clustering Algorithms and Support Vector Machines (SVMs) in the AI/ML space. Even One-Shot Imitation Learning (image recognition systems capable of learning new classes of visual objects from a single example, using probabilistic programming), which existed only as a theoretical concept and considered almost impossible, up until three years ago, has come of age and is witnessing implementation in practical applications.
- The amount of compute power used in the largest AI training runs has been increasing exponentially with a 3.5 month-doubling time (by comparison, Moore’s Law had an 18-month doubling period). Since 2012, this metric has grown by more than 300,000x.
- GANs or Generative Adversarial Networks, the technique of pitting two neural networks against each other in order to create a new piece of content – a picture or segment of music – is fast coming of age and is witnessing adoption. This technique elevates machines from being just learning entities to start creating and eventually evolve into beings with imagination.
- Reinforcement Learning, a third type of machine learning technique (apart from Supervised and Unsupervised), that uses evaluation and trial and error as learning methods (as opposed to instructions) is witnessing rapid advancements and implementations in robotics
- Third generation Blockchain technologies (such as NEO, IOTA, Stellar, Hashgraph, etc) built on DAG (Directed Acyclic Graph) technology that remove the miner from the equation are gaining popularity and are replacing Ethereum as a blockchain platform of choice since the beginning of 2018. While theoretically, these technologies are capable of handling hundreds of thousands of transactions (per second) as compared to 6 to 7 transactions per second by the (first gen) blockchain, we at Convergence Catalyst, have witnessed certain DAG based platforms perform over 10,000 transactions per second in real-world scenarios
- In the last couple of years, Internet of Things (IoT) has witnessed expeditious adoption in the industry and enterprise sectors primarily in the verticals of Smart Manufacturing, Industry 4.0, Smart Grids, Oil Rigs & Refineries, Wind Farms, Retail, Logistics, etc. Most of these industries have had sensors and been experimenting with sensor-enabled automation for a long time. Now with IoT, the focus is on Artificial Intelligence & Machine Learning, Security and Sensor Computing.
- Large technology companies are playing their part in the advancements of these emerging technologies – Apple’s Core ML 2 (a new version of its suite of machine learning apps for iOS devices) that the company launched at their developer conference in June, 2018 is 30% faster and reduces size of trained machine learning models by up to 75%. Amazon’s continuous improvement of Alexa’s ability to track context, intent and memory with the ability to interact with users for over 40,000 third-party skills and Google’s announcement of Duplex - the technology that carries out natural language conversations over the phone to complete real-world tasks, ML Kit – an SDK that allows developers to add AI to mobile and web apps, and third generation Tensor Processing Units (TPUs, Google’s custom accelerators for their machine learning applications), 8 times faster than previous year version earlier in May this year, indicate the advancements by the big, technology companies in the emerging technologies space.
Innovation and Disruption is Coming From All Corners of the World
While large, global technology companies are investing in advanced research in emerging technologies and integrating them into their own products and platforms, we have come across a number of small, young companies that are innovating and disrupting in this space:
- A one-member company that used the neural networks research in cancer genomics to develop an underwriting platform for insurance and financial services industries, raking in a million dollars in revenue from one client in one year
- A 2-member, US startup that vectorized a million pictures of food in less than two weeks using proprietary clustering algorithms and SVMs to build a “Shazam for Food” app
- A 3-member, Bangalore-based computer vision startup that cracked one-shot imitation learning technique to build a Video Analytics product for implementation in Retail industry… and, is using the same techniques and algorithm to develop a Sports Analytics platform
- A young, Bangalore-based company that is using advanced machine learning and “transfer learning” techniques to build and standardize machine-based local languages translation and transliteration in real-time
- An Indian fashion online marketplace that is working on GANs to design apparel for its house brand
- A young, Bangalore-based two-wheelers rental company that is cross-subsidizing its end-user rental costs through revenue generated from advanced analytics based insights provided to insurance firms to help them design customized riding pattern based insurance products and premiums
We are witnessing significant innovations and rapid advancements in emerging technologies from all parts of the world and companies of all sizes. This is being made possible due to the environment of open innovation and collaboration in the technology domain, especially the emerging technologies community. Most AI researchers share their work through whitepapers to the larger community, free of cost. Arxiv (Cornell University e-library) is one of most popular platform to publish one’s work in AI. Recently, thousands of AI researchers boycotted Nature Journal as they put some of whitepapers behind a paywall. Most of the companies developing blockchain based products, solutions and platforms document their architectures, methodologies and approaches in whitepapers for public viewing.
What used to be restricted to the domains of academia and corporate research labs is now being democratized. This is the collective Hive Mind at work! And, once one thinks through and comprehends this, the second half of statement made by the EmTech speaker – We are witnessing the slowest pace of advancements in tech for the rest of our lives – makes complete sense!
What Does This Mean For Companies and Industries?
Large companies with established business verticals and P&Ls that play the quarterly “appeasing the shareholders” game adopt a “Wait & Watch” policy when it comes to reacting to or adopting new technologies. And they cannot be blamed entirely. Some of these emerging technologies, especially AI and IoT have been at the peak of the hype cycles in the past and have undergone long winter spells between mid 1970s to mid 2000s. And the decision makers (senior management executives) of the large companies have lived through these buzzword-infused hype cycles in the past with little fruition, and hence have developed a strong bias against these technologies.
What is different this time is that the developments of these emerging technologies have garnered critical mass and speed, thanks to the coming of age and readiness of their enabler technologies (such as processing power, memory management, fast communication protocols, big data management, cloud, etc.) and the open, collaborative and transfer learning attitude of researchers and practitioners in these fields.
The management of large, established companies typically believe that they can use the power of money and influence to get back into the game at any stage, even if they miss the boat a few times. But, in today’s world of increasing public and private investments, perception and optics matter just as much as profits, and “Relevance” is a key metric of a company’s market cap. One example from our research reveals the case study of a fashion retail arm of an Indian business conglomerate, due to its old way of thinking and denial of adopting digital technologies based solutions across the board, is mired in with delays in inventory management. The stock arrives a full season late to the shelves in their retail stores, excess inventory of more than a year in the warehouses, poor vendor management… all these leading to the company’s inability to ever be able to turn profitable for the foreseeable future. On the other hand, a young competitor of this company, in the same industry segment purely by adopting Digital Transformation methods including implementation of batch processing, IoT sensors-based manufacturing and tracking, predictive analytics based demand forecasting, intelligent SKU management, analytics based pricing and quick turnarounds has managed to evolve as the leader (in terms of % profit) in the segment.
Another key trend we are observing is the faster and ready adoption of emerging technologies by non-core technology companies as compared to core technology companies. For a boutique consulting firm that has been (primarily) focusing on core technology companies for clients, Convergence Catalyst’s ‘Emerging Technologies’ practice witnessed a growth in projects from 2 to 11 in the past year (8 in AI/ML and 3 in Blockchain) across Travel, Urban Transport, Localization, Fund Management, Media, Cyber Security, Real Estate, Higher Education and Sports. This could be attributed to a combination of “living in denial” and “been there, heard that… Meh!” slumber mode of tech company executives.
In the telecom world, we had an axiom called “4-4-4 equation” to describe such scenarios – Telecom companies are made available solutions based on emerging technologies 4 years before the technologies hit the inflection point, telecom players (especially, telcos) take 4 years to whether to adopt these technologies and solutions or not? And, by the time, they do adopt and implement these solutions, they are 4 years too late. And, there are many examples of this axiom:
- Motorola realizing in 2008 that the innovation in the mobile phones’ space moved to software led by Symbian and Nokia, while it was busy innovating on hardware after the success of Moto RAZR in 2004
- Nokia realizing in 2012 that the innovation in smartphone space moved to design and UI/UX led by Apple’s iPhone, while it was busy with feature-led innovation for its operating system (Disclosure: I worked with the mobile devices arms of both Motorola and Nokia between 2005-08, and was privy to conversations both in and outside these companies)
- Telcos, globally, even to this day, trying to implement AI & ML based personalized plans, tariffs and products while the big data based analytics solution (under the label of “CEM – Customer Experience Management”) having been available since 2011
- Indian telcos realizing in 2016 that the innovation and, most importantly the 70:30 revenue share arrangement flipped and moved to Over-The-Top (OTT) players led by (among others) One97 Communications (parent company of PayTm), their erstwhile vendor
Technology companies are perfect examples of “Innovator’s Dilemma”. And, not just companies and industries, we are witnessing the early days of formation of tech communities concentration in different countries basis the local ecosystem, immigration policies, regulations (especially, for blockchain). Be it Estonia, Canada, South Korea, Israel, Japan, Eastern Europe or Indonesia, there are emerging technology communities concentrating in these regions driven by their support and open policies.
Having said that, not all companies are caught with their eye off the ball. Facebook, a less-than-fifteen year old company is moving from its core tenet of “connecting everyone on earth” to evolve as into a media company. And, this is due to its culture of “embracing change” and being constantly on the lookout for “Facebook Killers” and preempting them.
A Page From Facebook’s New Employee Handbook
Regardless of which industry a company operates in, the emerging technologies and their rapid advancements will impact every player – big and small – across geographies. And, companies cannot afford to “wait and watch” or “live in denial” anymore. They need to realize that the conventional playbooks are being thrown away and new ones are being written – The cheese is moving! They need to strive to be more open for innovation and collaboration. This is the moment to wear those running shoes and keep pace.
An abridged version of this Analyst Note was published as an Op-Ed in Mint's Quarterly Technology Review on 30th June, 2018.