VMI made a trip to San Jose, California, for the AI (Artificial Intelligence), Blockchain, and IoT (Internet of Things) Expo on November 29-30, 2017. We noticed 3 main trends in each of the segments, as well as some problems in search of a solution.
AI Expo Trends
- A move toward meta-learning in simulated environments – This shift helps AI to learn its mistakes away from potentially dangerous situations avoiding the open world for data collection—akin to Uber’s driving vehicles—and machine learning where “bad data” could teach bad habits.
- A new focus on distributed, open environments is becoming the best practice and path towards gaining high assurances in algorithms and more robust, secure environments for learning.
- Data no longer needs to be complete, correct, or perfect. Machine Learning algorithms running against excessive amounts of data – “Mega Big Data”- helps the AI system figure out on its own what is partial, incorrect, or irrelevant, which in turn builds more robust, informed systems.
Blockchain Expo Trends
- A major disruption to the Risk/Compliance industries through automated assurance from blockchain ledgers. Blockchain has the potential to completely shift auditing practices, allowing a constant audit to take place throughout an enterprise to ensure perfect balance sheets. GlaxoSmithKline and Viante are just some of the major companies already implementing the practice.
- A new approach to Cybersecurity through open-source transparency. Utilizing open-source architectures, this can allow an enterprise to maintain constant updates to its security policies by forking their cybersecurity issues to other enterprises, and learning from problems others report.
- A major shift away from fiat currency and rapid investment through new cryptocurrency vehicles. Bitcoin has hit $19,000 leading to a lot of excitement around the concepts of cryptocurrency as an investment vehicle, and the NASDAQ supported futures market speculated to help steady future volatility.
IoT Expo Trends
- A new approach to IIoT (Industrial Internet of Things) by interpreting data with Cognitive Analytics. This will give enterprise organizations the ability to learn from the IoT data they’ve gathered, analyze, and predict maintenance or other cyclical expectations.
- Data is in high supply and demand which can lead to a future of untapped Data Market Exchanges. As many companies are gathering data consistently and constantly, this leaves an asset that can be monetized on an open exchange.
- As companies expand and become more global, Telecom Connectivity is a more important aspect of a company’s ownership. This creates a need for encrypted management of devices across an enterprise which can control anything from bandwidth to territory.
While at AI/Blockchain/IoT Expos, we encountered many problems in search of a solution. These include:
- AI Experts, such as Dr. Danny Lange of Unity Technologies, believes that the data availability, algorithm development, and talent acquisition aren’t issues. The biggest issue AI faces today is the speed and reaction from management. Six-month cycles are now too long, we need six days.
- Going forward Artificial Intelligence needs to be able to explain its actions and how it learned its behaviors which lead to its recommendations. AI is an exponential force that is unable to give rationale to its actions currently.
- A developing issue with AI application. The knowledge is available in the industry, but there is no application: no cultural support, or no talent internally to find the connectivity from product to algorithm.
- Blockchain technology needs its “wow” moment for application. Currently surrounded by a flurry of activity and hype, the technology needs time to grow and be implemented away from its political and short-term investing background.
- In the world of IoT, data is still not seen as an asset to be monetized. This is causing many companies to lose out on alternative forms of revenue, as well as capturing entirely new areas of data.
- In the world of “Mega-Data” there are issues with “Dark Data” never being used, labeled, or tagged and therefor unusable in the context of analytics.