Neil Cornish explores AI today and in the future
Global forecasts for AI continue to be updated with overall revenues outstripping Cloud associated revenues over the next 10 years. According to Gartner, AI-derived business value is forecast to reach $3.9 trillion by 2022.
AI is not just another means of running a current Enterprise application, as with Cloud or Managed Services. AI is about fundamentally providing a new set of tools that will change how a business is run; how it’s products and services are developed and how the needs of the customer are understood and met. Language translation, news feeds, facial recognition, more accurate diagnosis of complex diseases, and accelerated drug discovery are just some of the applications on which companies are developing and deploying AI.
Successful participation in this market is not just about understanding what is possible today but understanding what could be possible. As timeframes for technology development come down, the challenge for many ideas will be in their delivery.
As algorithms continue to improve and indeed learn how to improve themselves, we are approaching an AI explosion. As AI seeps into every aspect of our lives, we will need to train more models. In the short term we will need more human help. We may need to Crowdfund the sourcing of AI trainers!
Industry watchers disagree on when this explosion will provide a SuperIntelligence, that superior to human intelligence, but it undoubtedly will. Reference Life 3.0, Max Tegmark.
As I presented at a recent DAMA conference, we believe that AI is where cloud was 10 years ago. Then cloud was something that small business put their website in or line-of-business used to get around the wait for IT delivering them a project. Cloud now has universal acceptance within IT and Business, underpinning many of the technology solutions delivered today.
Today AI is primarily being line-of-business driven with start-ups offering new solutions directly to budget holders. This will include marketing departments, HR and owners of manufacturing and research processes. Various parts of the healthcare system are also seeing AI as a point solution. There will also be many new market makers, however, their success will be about being first to sign contracts with major customers. Others will be close behind. This market will not take another 10 years to develop, by its very nature growth, like the complexity of algorithms, will be exponential.
Execution in some markets over the next 5 years may be harder than the solution design and development, because of the access to data and the restrictions imposed by Western Governments. Many of the constraints will slow down delivery whilst the likes of China will not be held back by concerns around individual’s freedoms.
For example, whilst China moves ahead with a central facial recognition system that can identify every citizen, enabling all sorts of follow on applications beneficial to the wider community, (crime fighting, payment systems, secure access), European governments are loading restrictions on what personal data can be retained, what it can be used for and for how long.
This is leading to some western developers taking their ideas to China. Access to tissue samples, for example, along with access to personal data, to help identify an individual’s propensity to react to certain cancer treatment strategies, is helpful to both the Chinese medical services and to the new business, as they test their AI application.
The challenge in the UK NHS of obtaining and managing a “Golden Record” for each individual citizen’s healthcare will require all those that hold data about us to share it. GP, dentist, physiotherapist, hospital, optician. And if you go into another hospital, dentist, optician do they create a new record? Do they get access to your existing record(s)? AI needs access to data. The more data the greater the insight and the increased probability of a successful outcome.
However, if these records are all shared, who has responsibility for maintaining them? Under GDPR this becomes more than a simple cost for each data owner. So how do you mine this data to help a patient that has come into A&E if you can’t see their history. A similar problem has been highlighted in a recent report in to the role of social services, health services and other government bodies played, with regard to a child murder case in Wales. The lack of data sharing was highlighted as a significant issue. If the data is not available, then AI can’t help.
The restrictions on data will see the development of departmental systems in much of Western Europe, as getting the wider buy in from Government or across large international businesses based in the West is too difficult. Meanwhile those nations “benefiting” from a benevolent, dictator style government will not be held back by such constraints.
Perhaps the most worrying development in the AI space is around autonomous weapon systems (AWS) development. With India recently announcing they are developing an AI based military strategy and the Chinese using AI to help them formulate their Foreign Policy and Diplomacy, AI will undoubtedly become part of developed countries’ defence systems.
The issue comes when these move from defensive systems to offensive. Or when these systems become available to those countries or individuals who have a grudge against another nation, or indeed all of us!
Google may have updated its AI Principles to exclude military use but many other companies and indeed countries will not be so principled. A global treaty on AWS development, similar to that for Atomic Weapons and Chemical Weapons, needs to be on the agenda.
There is so much AI can help us with, removing risk, extending life, diagnosing disease, that we need to embrace it. However, we need to ensure that the ultimate goals we set for AI systems are in line with our own. To do this, first we need to decide what our goals are.
This will not be a short debate and will make negotiating Brexit look like a walk in the park.
Neil Cornish - Neil has spent 37 years in the European IT Industry, working for IBM or as a Director of a Partner. Experience includes new business start-ups, acquisitions, disposals and management buyouts. Partnering with IBM until joining in 1989, Neil became UK AIX LOB Manager in 1991 followed by AIX EMEA Marketing Manager in Paris.
Later joining Anix Computers, the company built a reputation as the premier AIX Business in the UK. In 2007 Neil founded M7 Managed Services, an IBM partner, that launched an AI Centre of Competence in 2017, acquired by Meridian IT in 2018.
To find out more about Meridian’s AI offerings and partnerships, please join us at one of our forthcoming workshops. Click here for the next workshop.