Circuits Of Imagination

2018 Prognostications

23 Jan 2018

It’s the beginning of 2018 and over at KOPN Tech Radio it’s time for our yearly prognostications. Obviously no one can predict the future so take everything said below with a grain (or a tablespoon) of salt. With that said, here are my main predictions for 2018 and the future. Sadly reaching the singularity isn’t one of them.

Bitcoin Will Stagnate in 2018

Bitcoin has dropped by 50% or more at least three times in the last 4 years. In each case there was a rapid rise in prices followed by a drop and a long, slow downwards trend. In late 2017 early 2018 Bitcoin again experienced a surge in price followed by a drop. Based on previous trends in the price of Bitcoin will probably drop slowly over the next year or so, and then experience a slow but accelerating buildup before another spike and drop.

Bitcoin Price Chart

Of course Bitcoin could fail to ever deliver on its promise or regain it’s price. Both Stripe and Steam have ended support for Bitcoin (for now). On the technical side, Bitcoin’s maximum transaction rate has driven up transaction fees, making Bitcoin impractical for every day transactions and purchases. Also the continued growth of the Bitcoin blockchain (in terms of storage size) it self presents it’s own unique set of challenges.

Private (B2B) Blockchains Will Grow

Currencies are just one application of Blockchains. This technology is uniquely suitable for systems where there are multiple parties that do not necessarily trust each other but have the need for a shared and agreed upon record of events.

One such system is shipping. Tracking items between diverse shipping companies that do not have a full trust relationship between each other has created layers of signatures, verification, and bureaucratic overhead. A private blockchain shared between shipping companies could provide an easy and auditable cost saving opportunity, allowing each company to publish a signed record of an event (e.g. item received, item moved etc.) that can be checked and verified by the other parties involved. In fact just a couple of days after the show aired IBM & Maersk announced plans to create a global blockchain shipping venture.

Insurance is another area where B2B blockchains could have a positive impact on the industry. Simplifying the model and reducing costs can create new avenues for insurance. For example AXA is using a smart contract to manage flight insurance.Traditional insurance could also benefit from blockchain-based technologies, for example allowing car insurance companies to maintain policy status on a chain notifying the state if a policy has been cancelled.

There are also new initiatives to use the blockchain to track assets. The Canadian National Research Council is using the blockchain to publish funding and grant information, while the UN is using Ethereum to track and manage aid distribution to refugees.

All of these initiatives are still in the early stages, but the technology behind Bitcoin and Ethereum holds (hodls?) a lot of promise for changing the way businesses track, manage and transfer assets.

Tricking AI Will Become the new Code Execution

AI, Machine Learning and “smart” systems growth is accelerating in the industry and is starting to creep into different aspects of our lives. Traditionally, people have become used to computer algorithms determining what they see in their news feed and what they get when they search for specific terms. More recently, we are starting to see AI being used to predict the chance of committing a future crime and an AI has become the worlds first company director. In addition facial recognition has made huge advances to the point that Facebook can recognize people that aren’t looking at the camera and Google made amazing breakthroughs teaching an AI to play Go.

As AI and Machine Learning based systems gain more and more influence over our lives, manipulating these systems will become just as powerful as computer hacking is today. Already we are starting to see attempts to evade and fool Artificially Intelligent systems like the scarf that tricks facial recognition systems and glasses that trick facial recognition systems into thinking you are somebody else.

Perhaps most telling is the growing body of research on how to trick Neural Networks. For example, this easily accessible post describing how to trick an image classifier into thinking a panda is a vulture and other fun things. There is also a growing body of research on using generative adversarial networks to create inputs that create false classifications in Malware detection systems.

Wearable Personal Health Tracker

This one comes from one of my office mates, Tyler. We are already seeing Machine Learning effect the healthcare industry and this trend will continue to grow as Machine Learning techniques improve and Electronic Health Records become more accessible and standardized. Extrapolating from this trend, it is not too hard to imagine a wearable device that tracks individual health information and combines both long-term personal health data and population wide trends. This combined view can be used to generate future health predictions and create personalized diagnostics for emerging health issues; As well as alerting doctors to individual anomalies, trends and edge cases that can be used to form a personalized health care regiment.