Tuesday, January 26, 2016

Go Google Go! Why CogTech is the new frontier in learning

So Google (through an acquisition - Deepmind) stuck a camera in front of the notoriously difficult, 2500 year-old game GO, and it beat the European Champion 5-0. (It will take on the World Champion in March.) It uses a layered neural network with an executive or policy network to decide on each move. The clever thing was the training. The neural networks were trained on 30 million human moves then it started to play itself, using a trial-and-error process (reinforcement learning). This needed the gargantuan power of the Google Cloud Platform. This is a huge breakthrough as it shows that AI can take on much higher levels of difficult decision making tasks, those currently thought to be uniquely human.
AI has moved from theory to practice, as companies find a range of real applications for smart software. Almost all major tech companies reflect this in their AI investment strategies. As the Age of Algorithms is now upon us, we need to understand that this is a radical shift in the way we use technology. We are only just wakening up to the fact that it will transform the learning landscape. I have previously defined a 5-level Taxonomy for the use of CogTech in learning. There are already many examples of CogTech being used in the learning game, many more are on their way.
CogTech is technology that attempts to do things that minds do. This is not to say that the technology copies the mind, even the brain. We didn’t invent flight by copying the flapping wings of birds. It does mean technology that was traditionally the domain of the mind – that means many things that professionals used to do.
CogTech is many things
CogTech is not one thing, it is many things. Most of these things have relevance to teaching and learning. Computer vision can recognise images for search, even recognise faces for authentication in online assessment. Speech recognition can transcribe speech on YouTube videos and formatively assess in language learning. NLP (Natural Language Processing) can be used to automatically create online learning and help automate essay marking. Optimisation can be used to deliver personalised learning to learners. Adaptive systems can ‘satnav’ learners through learning experiences, based on individual and aggregated data. Machine learning can be used in learning systems to improve their own performance. Even robotics can be used to assist in teaching. All of the above are being used NOW.
These technologies have the ability to transform online learning from the current era of flat, linear delivery to smart, inferred, feedback-rich, personalised, predictive, data-driven delivery. In short, the technology will increasingly deliver good teaching and learning.
CogTech is huge
CogTech has already influenced many sectors; search, finance, sports, online dating, entertainment, health, retail and so on. The company to watch here is IBM. They have invested huge sums in CogTech and have a strategic intent, through Watson, to offer cloud-based intelligence. Healthcare is their big target through their Watson Health Cloud Platform. Then there is Google who have made over 20 CogTech acquisitions, including Deepmind, in just three years. This is a huge strategic move under their Alphabet restructure. It sees them move from the old Darwinian model of ‘lots of projects in the hope that some will stick’, to more focus on smart CogTech.  Amazon, as an enhancement to Amazon Web Services has launched Amazon Machine Services. AWS is huge and may in the future dwarf their retail business but it will only do so if they innovate through machine intelligence, that they are doing. Microsoft, of course, has added Microsoft Azure Machine Learning to their Azure platform. They have to move away from their traditional PC and operating system business and they have. Facebook, Apple, SalesForce, Oracle, Cisco and Intel have also been building and investing in CogTech.
CogTech & learning sector
The learning sector (education and training), for me, is ripe for a radical boost through CogTech. It is, after all, a sector that is NOTHING BUT the delivery of cognitive improvement. Using Cognitive technology is an almost perfect fit. It is also a sector with increasing demand, not only in numbers but also quality. It has high labour costs, and the current labour-intensive models have plateaued in terms of productivity. Huge sums have been invested in the developed world in the current, largely classroom/lecture hall model, with no concomitant increase in attainment. The model is crying out for scalable solutions, not on fixed, linear teaching and learning but on smart teaching and smart learning. Resistance is futile but reasonable and organised assimilation is desirable.

No comments: