Employment: Blockchain developer is the fastest growing job category

What: According to a survey of 1,000 hiring managers by freelance marketplace Upwork, the fastest growing job category is ‘blockchain developer’. The number of jobs grew 6,000% in quarter 1 of 2018 [1]. On the other hand, blockchain developers ranked only seventh in the skills most desired by hiring managers, with Big Data and Analytics topping the list [2].
Implications: While digital currencies such as Bitcoin and Ethereum may go through boom and bust cycles, it appears as though enterprises are investing heavily in blockchain technology as a trusted way of exchanging transactions with third parties.
[1] Computerworld
[2] Upwork

Artificial Intelligence: AI benchmarking shows huge improvements

What: Stanford University has established project DAWN to benchmark the performance of deep learning systems. DAWN is basically a competition as it uses a benchmarking tool (DAWNBench) to measure the performance of solutions presented by different teams for both training and inference. They have just released the first DAWNBench report which shows that in the last six months training times for the ImageNet (a database of images with objects to be recognised [1]) challenge have come down by almost 500-fold and inference latency has dropped from about 50mSec to 10mSec. In addition, the cost of inference has dropped six fold [2].
Implications: There is a new industry developing in benchmarking for AI systems. Another initiative supported by Google, Baidu, Stanford and others, MLPerf, has somewhat similar but different aims in that it benchmarks AI chip performance as well as software systems [3].
[1] ImageNet
[2] Stanford
[3] The Register

Mobile devices: Multi-touch skin

What: A German research group has come up with a novel way in which to use your skin as a touch sensitive input for a mobile device. To be more accurate, it is actually a sticker which you attach to your skin and allows multi-touch input like a touch screen. While the device is still an experimental prototype, the researchers have shown that you can use the stickers to send a sms, control a phone call or manage a music player [1].
Implications: At some point in the future we shall move away from the glass slab form factor of today’s mobile phones and our devices will transform themselves into an array of wearable devices such as a watch with the processor and cellular radio, and a heads-up display built into glasses. However, we would still need some mechanism equivalent to today’s touch screen to enable data to be input. While voice will be a suitable alternative in some circumstances, people are also experimenting with gesture inputs because most of your input will need to be done in silence.
[1] Engadget

Telecommunications: Google’s latest messaging app

What: Google is renowned for the number of attempts it has made to produce a decent messaging app: Hangouts, Duo, Allo and Messages. Other than the latter (which is basically just a SMS app) the other three are confusing and having overlapping functionality and are basically failures. Now Google is about to have another go with an app called Chat which (I think) will be inbuilt as the default messaging app in Android. Chat uses RCS (Rich Communications Services), a protocol defined by the telcos for multimedia messaging and presence. It has been around for over a decade and gone nowhere. Nevertheless, Google has invested a great deal of time and effort in signing up a batch of global carriers and getting them to conform to a common RCS profile. They have also put one of their top executives (the guy who was responsible for Google Photos) in charge [1].
Implications: It is hard to understand why Google think Chat can succeed. Unlike the competition (iMessage, Whatsapp, Facebook Messenger), RCS is a carrier service, so carriers need to invest in servers to manage it without any clear revenue opportunity. Surprisingly, a number of carriers have done this. However, unless Apple support RCS – an extremely unlikely proposition – it is hard to see anyone being excited by a service which serves (in Australia at least) less than half the market. Chat will default to just being a new SMS app.
[1] The Verge

Energy: NASA tests small portable nuclear reactor

What: Just when everyone thought nuclear power was dead, it has been reborn – as a baby. NASA has just conducted tests on the first new fission nuclear reactor design in 40 years. The goal of is to have a power unit that is small and light enough to be transported to Mars for an extended stay there (the fuel for conventional power sources would be too heavy). The reactor, known as ‘Kilopower’ ranges in size from 1KW to 10KW and is intended to operate without human maintenance. NASA tested the reactor in Nevada to try and simulate extreme conditions and found that even with the cooling system cut off the core temperature did not rise more than 15 degrees [1].
Implications: If nuclear fission power has a future it is in small modular reactors such as these. However, there is no indication of whether these would be cost effective for anything other than a highly specialised purpose such as a Mars mission
[1] Engadget

Energy: Progress in battery technology

What: The capacity of a lithium ion battery is determined by how many lithium ions can be attached to the anode during charging. It is well known that replacing the carbon anode in a lithium ion battery with silicon would increase the energy density because silicon can capture 25 times more lithium ions than graphite. However, to date no one has found a way of making a silicon anode that does not expand and crumble when the battery is charged. However, a startup called Sila Nanotechnologies, founded by ex-Tesla battery engineers, emerged from stealth recently to announce a partnership with BMW to develop new batteries with silicon-based anodes. Sila have used materials science expertise to create a new type of silicon nanoparticle for an anode which they say will initially increase battery capacity by 10% to 15% and ultimately by 40%, without expanding or crumbling. Because of the long cycle time for car components BMW will only deploy the technology in vehicles in 2023, but Sila say that consumer devices with the new technology will appear as early as next year.
Implications: Energy density is only one critical parameter for batteries. Sila say their batteries will have “normal performance and cycle life” compared to today’s lithium batteries, but there is no mention of cost.
[1] MIT Technology Review

Healthcare: Google brings Augmented Reality to the microscope

What: Detecting cancerous cells from biopsy tissue requires a cytologist to laboriously scan the cells under a microscope. The slides contain a vast number of cells, each of which needs to be examined so it is a time consuming and labour-intensive process. Google has come up with a novel solution which is to attach an augmented reality (AR) system to an existing optical microscope. The system feeds the images into a computer where a machine learning algorithm scans the cells to detect the ones of interest. The AR system then marks up the slides in the microscope directing the cytologist’s attention to the highest priority cells [1].
Implications: This is a novel intermediate step between manually intensive activity and full automation. While it does not replace cytologists it makes them much more productive, providing the machine learning algorithm does not miss anything.
[1] Venturebeat

Transport: Open source test software for autonomous vehicles

What: Autonomous vehicles will incorporate complex technology and a comprehensive test system will be critical to ensure safety. A US company, Voyage, which deploys self-driving taxis, has developed a range of software test tools which it is open sourcing for all manufacturers to use. The software will enable users to develop test scenarios, inject errors into the system and determine the impact of technology changes on the overall system [1].
Implications: By open sourcing their solution, Voyage are enabling all developers to contribute to what could become the “Linux” of the self-driving world.
[1] Voyage

Healthcare: Detecting brain disorders with your computer or smartphone

What: A number of startups are testing whether a change in the rhythm of your use of your computer or smartphone could indicate the onset of conditions such as Parkinson’s, Alzheimer’s or depression. Silicon Valley startup Mindstrong has collected data on people’s typing and scrolling on their smartphones via an app on the phone, in order to detect depression or schizophrenia. The app is being used by drug companies to determine the efficacy of medication for these conditions.  Another Silicon Valley startup, NeuraMetrix, is hoping to use the fact that our typing style is unique to detect Parkinson’s, depression or Alzheimer’s. The key is not the actual way in which we type but any changes that are detected from day to day. Toronto based WinterLight Labs is drawing similar inferences from speech parameters [1].
Implications: Non-intrusive measurements that extract data from our daily routines show great promise for the future of healthcare in that they allow large data sets to be collected at minimal cost.
[1] IEEE Spectrum

Machine vision: A new object recognition app from a school student

What: According to the Imagenet Challenge [1], machines are now able to classify objects in a photograph more accurately than humans. However, to date even Internet giants such as Google have failed to produce a decent smartphone app that will accurately recognise objects you point at (not for a lack of trying). TechCrunch, however, reports that a school student, Michael Royzen, has developed an app called Smartlens which does a reasonable job at recognising up to 17,000 objects. The app uses convolutional neural networks and does it all locally in the phone without cloud lookups. However, that does necessitate a 500Mbyte download onto the phone [2].
Implications:  An object recognition app could have a significant impact on the retail trade by providing customers a simple way to find suppliers for something they are interested in. However, it is not clear what the threshold accuracy rate would be for user acceptance and whether SmartLens actually achieves this.
[1] Kaggle
[2] TechCrunch

Professor Hugh Bradlow is President of the Australian Academy of Technology and Engineering. He is recognised as a global leader in telecommunications technology, including being named by Global Telecom’s Business in the top 100 most powerful executives in the global telecoms Industry two years in a row, and by Smart Company as one of the 12 most influential people in Australian ICT.