There was considerable hype around AI over the last decade. Initially there was a view that AI would be able to replicate human intelligence. DeepMind, the AI lab purchased by Google, espoused the concept of AGI (Artificial General Intelligence) which would rival the human mind. leading philosophers and futurists painted a picture where computers and robots powered by AGI would rule our lives……………………………….
As the decade progressed the capabilities and limitations of AI became apparent. AI provided helpful outcomes for specific applications; hence the term narrow AI. Examples include, face and image recognition, sentiment analysis, predicting buying behaviour and controlling robots and driverless vehicles.
Technical issues, such as lack of affordable computational power and data storage have been overcome, using cloud-based services and GPU powered computers. The usual limitation in developing effective AI solutions is now the availability of enough consistently labelled data.
As well as the usual challenges faced by new technology; the human aptitude for change and data security and privacy, AI has another hurdle to cross – trust. AI may give biased results and can be used to generate fake information.
Looking into the 2020’s:
- I believe that organisations will invest in managing their data assets and hiring people with skills to derive insight from this data. I think it will be not until the middle of the decade that we see the establishment of the Chief Data Office as an important organisational department.
- Governments and business should do more, to encourage people to think about how they can use AI to augment their daily work lives. Government policy should encourage organisations to help their staff develop the skills they require, so they don’t end up becoming a redundancy statistic due to automation.
- I hope that the tech giants and policy makers worldwide can collaborate to develop standardised models, to mitigate the pervasive privacy concerns associated with the Internet and latest communications technologies.
- Individuals should become much more discerning when they peruse content to decide on its validity. Governments can facilitate this with awareness campaigns, such as those we see in the UK to counter telephone scams and by outlawing undesirable activities.
AI offers us the opportunity to address some of the most pressing productivity challenges facing our society. At the start of this year, Nature published a paper by researchers from Google and Imperial College London. This outlined the extensive research they had undertaken into the use of AI for breast cancer screening. The paper indicated that AI was just as effective as the current system where two radiologists look as each sample and better than a single clinician.
I understand that the UK NHS is rightly (in my opinion) not considering using AI to replace human input, but in using it to augment the clinicians’ modus operandi.
My thought as we enter the new decade, is that we all need to think more about using AI to augment human capabilities rather than replace them. For advice on developing a practical approach to using AI please get in touch.