Annie Bailey, Senior Analyst at KuppingerCole, will host the Artificial Intelligence stream at European Identity and Cloud Conference 2021.
To give you a sneak preview of what to expect, we asked Annie some questions about the future of AI.
What is AI's future?
We can think about the future of AI in three different ways. So we could of course think about it from the technological advancements, which are happening right now, and really speeding up at a rate that we haven't seen before. So there are advancements in AI in conversation and cognitive use cases, towards analytical use cases, towards integration and automation for all sorts of moving things: into cars, into robots, into drones, and then of course connecting all of these together in the processing side of these. In just about any industry you can think of, there are really strong and sometimes surprising advancements here in AI. And so that future is really driving forward at a rate that's quite exciting. We could think about the future AI from the regulations standpoint too, and that's really going to have a huge effect on what AI will become and how it's going to be integrated into our businesses and our lives.
There are of course, many different frameworks around the world. There's not all that much legally binding regulation that's out there, but on the horizon, there is a recommendation for a regulation which has been put out by the European commission. We'll talk about that a bit more in depth later. We can think about the future of AI as it's being shaped by business. There are so many use cases here that are really working towards the process optimization side and assisting in decision-making, assisting in transferring information into a way that can support fast decision-making. So these three things are really creating the future of AI - so we could say technology, their regulations and its applications in business.
What do the regulations indicate about AI's future?
If we consider the journey that AI has been on and the journey that regulation has been on concerning AI, it's gone from frameworks and ethical statements about how AI should relate to and be a part of human society to where we are now. And today, there is a clear cut proposal for regulation from the European commission, and this could be adopted sometime in the next year to two years, so we're of course not in the place where we have legally binding regulation in the European union yet, but it's on the table. If we think about what this regulation is actually trying to do, it's working to harmonize the use of AI, and its development across the European union. So it's very typical of EU regulations. There needs to be a standard and interoperability between all member states.
What's quite interesting is they go into a tier based system on the risk that any AI system could potentially give to any of its users or any stakeholders around it. It breaks it down into regulations for systems, which have an unacceptable risk and how to deal with those, and what counts as an unacceptable risk and then down to high risk, limited risk, and then finally to minimal risk. And so we're moving towards a little more clear cut understanding of how AI is going to be used in our societies, and what's going to be acceptable and not, and what the penalties for that will be. So that's obviously very good in terms of removing uncertainty, but that's of course, laying down some boundaries, and with boundaries its going to bring oversight, need for governance and accountability here. So both sides of this you can argue are going to swing wide the gates without so much uncertainty, the future of AI as much brighter because it's clear what is acceptable and what is not. On the other hand, [there’s] a more limited take on the future of AI, because of course there are horizons and boundaries, which should not be crossed.
How are enterprises getting AI into their business?
Businesses have a few options for bringing AI into their systems and processes. The first could be point solutions, and these are AI systems that have been developed and trained to do one very clear, specific job, or to meet one very clear, specific use case. This could be anything from the health industry detecting malignant tumors in images and scans to supply chain management, to anomaly detection for a specific use case. These are all wonderful and great and meet a very specific and defined need, but of course, if you think about the spread of companies, which will need to be developing these solutions, it's endless and every situation, every scenario that a company may encounter, would like AI to solve that problem or to step in or to support these uniqueness factors are so great that we're never going to come to the end of different point solutions.
Another option has turned up in what we're calling AI service clouds. Now these are cloud solutions that deliver the whole range of machine learning development and governance tools, as well as the computing power and the cloud power behind it. These options tend to be really good for organizations which want to create their own point solution for their own scenario, and so these are becoming quite interesting. AI is being brought into the product itself to help automate and structure the process of selecting a model and going through training and validation, then through the lifecycle management and the ML ops [Machine Learning Operations] side. With that, businesses are able to either go for the point solution, something which is already designed for a very clear, specific use case, or they can be a bit more creative and design it themselves with the support of an AI service cloud.
Flip the script, what does AI mean for our future?
We've been talking about what the future of AI is, and of course what's impacting that, but we have to think about this as a relationship. What does the impact of AI mean for our future? And for us, it's going to mean a much higher dependence on data, and we're already in this data-driven world. Every day it's bringing us closer to having more data-driven insights, to being able to collect data from across the organization to break down silos and use this in really smart ways. AI is going to accelerate that and bring quite measurable benefits from that, which is only going to speed up the process. With that higher dependence on data, we have to have more resilient data value chains and governance around data. This is absolutely essential and will be even more so as we speed up this process.
And of course, AI is many, many things, and it's such an ambiguous term, but you can think of it as a form of communication because AI is bringing information into a form which can be used to make decisions. And so that's a form of communication, or of course you have the natural language aspects where you're being able to communicate your needs in a more natural way, rather than coding, querying or just using keyword searches. All of these methods of communication and bringing communication more into our daily business processes is going to bring to light the need for transparency. There's going to be more opportunities to ask the questions of how was that data collected, or why was that decision made, or why does this impact me in the way that it does? Those aspects, the dependence on data, the need for then governance and security of that data, and then transparency into our heightened ability to communicate, are going to be some outcomes of AI in our lives.