Posted Wed 17 Oct 2018
Reality: AI is a general term which describes any technology that performs tasks which demonstrate characteristics of human intelligence. Machine learning is a subset of AI which can achieve ‘intelligence’ with the use of algorithms to ‘learn’ with data, without being explicitly programmed. Deep learning is a subset of machine learning based on learning data representations, contrary to task-specific algorithms.
Reality: Research has been going on for 60+ years (becoming an academic discipline in 1956) though recent media coverage of the new applications and developments of AI may give an alternative impression.
Reality: The current state of AI is still in its early stages, subsequently AI needs to be engineered and trained before being applied to a problem. However, AI has a notable collection of models in sensing the world e.g. speech recognition, though even these are unlikely to be a one size-fits-all in the case of sensing the world.
Reality: Gartner (2017) predicts that although 1.8 million jobs could be lost to AI by 2020 they predict 2.3 million new jobs could also be created as a result. This ‘creative destruction’ has been seen throughout time with the development of new innovations - for example, the industrial revolution shifting jobs away from agriculture to factories. It is more likely that AI will transform the employment landscape more in terms of ts composition, as opposed to solely raising unemployment rates.
Reality: AI-based systems rely heavily on humans in order to gather data about how humans carry out activities they monitor - hence there will always be a need for humans. Though once established some AIs may be able to operate autonomously and independent of human supervision, this is unlikely to be the case in key areas or critical systems.
Reality: 17/18 banks already apply some form of AI in their front office (Centre for Finance, Technology and Entrepreneurship, 2017).
Reality: applications of AI in finance include but are not limited to: data collection; enhancing customer personalisation e.g. with robot advisors in wealth management which can automate the rebalancing of portfolios based at <100 base points compared to traditional brokers operating a 200-300 base points (The Financial Brand, 2017); and fraud detection givens it ability to constantly evolve to threat changes through continual learning