A plain-speaking guide to AI and advanced analytics: A free resource to help build trust in a technology crucial for sustainability

It is highly likely we will need AI to help address existential needs, such as sustainable solutions to the climate emergency. But without trust in AI - and technology more generally - these solutions are likely to be delayed or even abandoned.

My Fujitsu colleagues/co-authors and I talked about this and decided the time is right for a grounded and rounded view of where AI and advanced analytics are today. One that does not talk down to the novice and does not over-promise, over-claim, or paper-over the areas where AI is currently weak. That whitepaper has just been published and is now available as a free resource. A link to the whitepaper is available at the bottom of this page.

We have aimed to cover all the essential aspects of AI, leaving you to pick your passion and drill down into the information you need. We cover the technical aspects of AI and advanced analytics, but we also put this into context in terms of how they are driving change in business and our day-to-day lives. The information will enable you to transition from novice - if that is where you are today - to enthusiast.

The paper starts with some examples of how AI has been used over the last decade. My favorite early example comes from 2011. Conservationists researching the implications of bird collisions with powerlines realized they were overwhelmed by the amount of audio data collected from power line stations, plus the associated metadata, such as times and locations. They turned to an AI specialist and were able to detect collisions automatically. Results suggested that bird deaths were actually much higher than expected. This has resulted in new, evidence-based conservation approaches.

That takes us on to a discussion of the different types of AI. The first classification is by capabilities. These are task-based narrow-AI, self-learning general-AI, and super-intelligence - the so-called "singularity" - that would be the most capable form of intelligence on Earth.

The second classification is by functionalities. There are multiple types of AI systems:

• Reactive Machines are the oldest and most limited forms of AI systems. These emulate the human mind's ability to respond to different kinds of stimuli.
• Limited memory machines. Nearly all existing applications, including deep learning - come under this category, and learn from historical data to make decisions and predictions.
• Theory of mind AI. Today, these systems are, at best, works in progress. Eventually they will be able to discern the needs, emotions, beliefs and thought processes of people and other AI entities they engage with.
• Self-aware AI - the final stage of AI development. Today, this exists only hypothetically - and is expected to occur when AI has evolved to have what philosophers call "the hard problem" of attaining consciousness.