January, as well as ushering in a new year is a time of predictions for the upcoming 12 months. The world of Machine Learning (ML) is not immune to this, though as the predictions are generated by humans and not machines, they must be taken with a generous pinch of salt.
Tech history is littered with false dawns, crushed hopes and wildly inaccurate pronouncements about the future; two of the more infamous being:
“I think there is a world market for maybe five computers.“
Thomas Watson, president of IBM, 1943
“Two years from now, spam will be solved.“
Bill Gates, founder of Microsoft, 2004
There is no consensus on what 2019 will bring either for Artificial Intelligence (AI) or ML, however there are two general strands of thought from the experts in the field which can be broadly characterised as evangelical or cautionary in nature.
The ML evangelists concur that 2019 will be the year in which ML expands exponentially into almost all areas of work and leisure.
Most agree that data science will be at the crux of this development; predicting a large growth in the data available, progress in methods of acquiring, storing and accessing it and an increase in the relevance of data managers, experts and developers.
This is viewed as an essential pre-cursor to a similarly predicted expansion in the use of ML in a range of areas, from virtual assistants & chatbots to retail & commerce as well as in the implementation of governmental initiatives & policies.
Machine Learning is also viewed as critical in the ongoing evolution of AI development. Manoj Saxena in a Forbes interview further suggests that realising the potential of AI will be a 500-year endeavour, not a 50 year one, emphasising the importance not only of ML, but also neural networks and deep learning.
Another predicted trend is for an increase in and development of the interaction between humans and machines. This is viewed as essential if the full potential of ML (and AI generally) are going to be realised. This has also been described as the need for a social contract between humans and machines.
More cautionary predictions are also being made this January; the year head is expected to provide some challenges for ML.
There is an increasing awareness of the encroachment of ML and AI into aspects of life that until recently have been considered impervious. The use of data science in recent elections worldwide to attempt to influence opinion, voters and results has been brought to the fore. This goes hand in hand with the deployment of similar technology by governments to control, stifle or shape public policy, opinion and debate.
Similar concerns have been raised about the nascent nature of ML, in that it can be deceived at best or exploited at worst by relatively simple alterations to the data on which it is working. There is also a growing awareness that this can work both ways; that ML can be unscrupulously used to deceive people.
Furthermore, whilst most forecasters predict an increase in the use of ML in retail, recent surveys have suggested that there is considerable push-back from consumers against this. This is most pronounced in the reactions of consumers to chatbots, with some marketing experts stating that this will hinder public uptake and engagement with new technologies.
There are, as ever, the wilder predictions. These include predictions that the tech in Blade Runner (set in 2019) may well be appearing this year, from the optimistic use of AI and ML in voice driven photo editing, to the slightly more alarming development of functional replicants.
Less worrying are suggestions that robots will play an increasing part in domestic life, with some suggesting that they will become part of our families, take on dull and dangerous jobs and even act as care-givers.
HappAI New Year.