Earlier this week, we were delighted to host a webinar led by IBM’s Global Chief Data Officer, and CDO of the Year, Inderpal Bhandari. During the broadcast, Inderpal tackled the topic of ‘What Can AI do for You?’ and explored how Artificial Intelligence can revolutionize your enterprise.
The webinar was the first of two in the run up to the IBM CDO Summit. The Spring Summit takes place on May 1-2 in San Francisco and will bring together over 120 CDOs, CAOs, and other C-suite & VP-Level execs responsible for data and analytics. Be sure to apply for your free pass to join them.
The next webinar in the series will feature Allen Crane (AVP, Applied Analytics, USAA) and Caitlin Halferty (Client Engagement Executive, Chief Data Office, IBM). Entitled ‘The Holy Grail – Exploiting your Data Assets to Deliver Real Value’, the session will take place on April 24th at 8am Pacific / 11am Eastern.
In case you missed this week’s webinar with Inderpal, here’s a recap of the key takeaways from the session.
Takeaway 1: What really counts as an Artificial Intelligence system?
Inderpal explained that he defines AI systems as those which have 4 main attributes, which he refers to as the 4 E’s of cognitive:
- Expertise: The primary value of an AI system is that it must be an expert at something – whether it’s a process (e.g. supply chain management) or a domain (e.g. oncology), this expert knowledge is what gives the AI system its purpose.
- Expression: AI systems must support natural forms of expression, so that they can interact with a human. AI systems are used to support better human decision making, and so there will always be a human in the loop, and the system must be able to not only explain its recommendation, but also the reasoning behind it. This is especially true in the business enviorment when decisions could have a major impact of the enterprise, and so if it is a black box, the recommendation will be ignored.
- Education: AI systems are educated, not programmed. This education is performed by giving examples, providing input, and entering new data. It is a process in which the user is likely to learn as much from the system, as the system does from the user.
- Evolve: They are able to change as they learn more from the data inputted, and they can do this at scale. One reason AI systems are becoming increasingly important is that they address the exploision of data. By 2020, medical knowledge will double every 72 days. Even today, Doctors would need to spend 160 hours a week to read all the new literature that is being produced. Therefore, the AI systems ability to evolve at scale to process all this information is crucial.
Takeaway 2: How can Enterprises make best use of AI technology, to become ‘thinking businesses’
Within an enterprise, there are 1000s of subject matter experts doing their job. To become a ‘cognitive enterprise’ or ‘thinking business’ all this means to be put an intelligent system into the loop with all of these subject matter experts, to supplement and improve their decision making abilities – whether that decision is big or small, AI can be included.
IBM has set out to become an AI driven enterprise, and so provides an interesting model. At IBM, all processes now include an intelligent system within them. The subject matter experts work with AI to identify what data is relevant (structured and unstructured, as well as internal and external data), how to source it, set the context, and then move to generate insight to underpin the decision made.
As all processes loop the same intelligent system in, the system becomes the hub, and this leads to much greater efficiency. Whereas before, decisions would need to be escalated, or move laterally to other departments, with everyone working off the same system, this means that all the information and insights are in one place, and so the processes are dramatically sped up. The system holds much more information than a single person ever could, and coupled with its ability to capture real time information; it is able to support much more accurate decisions.
So what is an AI business? Inderpal explained that to him, they all have one thing in common – that is putting the same intelligent system at the heart of all their processes, and as such, they see a dramatic reduction in cycle time in those processes. The cycle time can in fact be reduced by several orders of magnitude.
As you can see, the AI enterprise requires an interaction of four key component; data, technology, business processes, and organizational considerations. These four key components underpin Inderpal’s AI Enterprise Blueprint, which maps out the target state to fully exploit AI systems within your organization.
Takeaway 3: You need an AI Enterprise Data Architecture to underpin your AI system
Of course, to get to their current state where AI is included in every process at IBM, a lot of work was required. The first step though, is to ensure that your Data Architecture is set up to handle the requirements of this goal.
Inderpal encouraged attendees to think of this new enterprise data architecture as building the data & AI backbone for the company.
There are several specific requirements for this type of architecture, of course there will be variation in different organizations, but some of the common themes, challenges and requirements are:
- Must be enterprise wide, so that all information is in one place and the system can act as a single hub or backbone. Whilst this is a significant engineering challenge, if every process had to create a separate data lake it would be incredibility hard, and also would lack the interaction between different parts of your organization to truly be AI driven.
- Achieving the above will also provide the enterprise with a single version of the truth, which is trusted, and certified, leading to genuine consistency at the enterprise scale.
- A single platform that provides both the data storage, and the compute nodes – at least to appear this way to its users.
- A need for not just structured data, but unstructured data also
- Needs very fast interaction between the lake and other systems, due to the need for it to provide real time info for systems using tech such as deep learning.
- A requirement to move data at scale, very fast. IBM made the decision to put their whole data lake on flash memory for this reason.
To find out more about AI technology, becoming a data-driven enterprise, and much more, be sure to join the IBM CDO Summit in San Francisco this May 1-2, and if you’re interested in extracting value from your data, be sure to register for the next webinar on ‘The Holy Grail – Exploiting your Data Assets to Deliver Real Value’ with Allen Crane (AVP, Applied Analytics, USAA) and Caitlin Halferty (Client Engagement Executive, Chief Data Office, IBM)