Machine learning, neural networks, artificial intelligence, companies have long been using these revolutionary technologies for specific tasks. But can an AI model be used to manage an entire corporation?
Companies consist of several organisational units that function autonomously and yet represent a larger whole. As a result, a great deal of data is generated in multiple locations. In many cases, the other internal organisations have been unable to access this output, or cross-correlate it in a meaningful manner, says Bill Wright, head of AI/ML and intelligent edge, global industries and accounts, Red Hat.
Also, the evaluation of all this data by humans is time-consuming and costly, and some companies are using dedicated AI models to support them. But given the lack of integration, the larger optimisation and trend analysis benefits are being lost in the process.
The synergistic effect between the individual business areas will be far greater if this wealth of data and the corresponding AI models are combined in a single system for analysis, and the observations and conclusions can be presented and made available to all related employees. Autonomous actions can be cross-checked for better impact and accuracy. And additionally, there can be a significant reduction in the error rate when humans are supported by machines to recognise, evaluate and consistently use complex correlations with regard to business decisions.
Enterprise neurosystems: a new industry-wide initiative
Four years ago, we began a focus group at Red Hat to look at the intersection of AI and networking. We discussed potential architectures and uncovered some of the deeper issues facing our customers. They were making a significant investment in AI models for point solutions, but there was no overarching framework for integration and cross-correlation. Model sprawl was taking place, and collective analysis was missing, so this quickly became our area of focus.
To answer these challenges, Red Hat collaborated with a number of tier 1 telecoms operators and top academic institutions.
The goal, to design a new fully integrated AI framework, similar to a human neurological architecture. In this old yet new paradigm, elemental functions would be managed autonomously, and others would require a level of consciousness with wide-scale analysis and predictive capabilities. This framework is intended to provide the missing infrastructure to link existing AI models and data sources together and create a collective intelligence. In this way, Edge AI and Core AI will unify and become one centrally managed intelligence system.
This is another step towards the end state, a Business Singularity for every company. An intelligence that can act as a guide and collaborator for the human members of the organisation. It is also a framework that can work with other firms’ intelligence systems, to streamline processes and achieve mutually beneficial objectives.
As mentioned, internal departments still create AI models on an individual basis, depending on the need and feasibility for smaller areas of the business. Alternatively, certain providers offer them off the shelf. Seamless integration and a central interpretive/reporting methodology across all these instances is missing.
The Enterprise Neurosystem is designed to remedy this situation. Connected to all business units within a company, it acts as a central cross-correlation engine, pulling insights from all the machine learning technologies already in use. It takes data out of all silos of operation, both real-time and historical, to compose an integrated view of company operations in a unified manner. This wide-ranging implementation ensures deep insight into the data sets of different departments, and how they impact each other over time. This enables the algorithms to establish correlations and cross-connections of previously unattainable predictive value.
Intelligent middleware will also be a key function and focus of the community. We will develop an autonomous capability that can seamlessly detect and connect to a variety of AI models, applications and data repositories. This in turn will enable the overall use case, and contribute to ease of deployment.
The interface between this AI framework and humans is initially designed as a classic dashboard. However, the community’s objectives are going much further, in the direction of a holographic assistant that acts as an advisory colleague, and offers advice on both real-time issues and options for course correction.
One of the community’s working group tracks will cover Ethics and Governance. This will imbue the core intelligence with humanistic best practices, and incorporate ethical guidelines from a wide array of cultures and philosophies. This will establish clear baselines and boundaries of ethical behaviour for the core intelligence to follow, as it will make business decisions and deliver recommendations that will not only impact employees, customers and partners, but the environment itself.
Technical implementation
Structurally, an enterprise architecture could be composed as follows, The foundation is the requisite hardware with state-of-the-art software platforms, such as Red Hat Enterprise Linux, Ceph, Kubernetes and Data Streaming. On top of this is an open data framework that allows the use of open source AI platform tooling. Employees use the latter to create AI models, put them into production and manage them. The beauty is that both the models and the tooling required to maintain them are already available.
Hovering over everything is the Enterprise Neurosystem. It autonomously manages many elements of corporate IT and logistics, the same way the human body manages primary functions (the heartbeat and so on). But once fully implemented across the company, this system establishes valuable cross-connections to all data within the organisation. Like the conscious mind, it evaluates and establishes current/future patterns, and after the requisite training, makes optimisation suggestions.
One way to look at it is that business intelligence will now be fully integrated with AIOps in a flexible, self aware architecture. That architecture is the Enterprise Neurosystem.
Open source as a driving force
Open source plays a crucial role in this initiative. A wide variety of verticals can adopt this technology platform, and insights from all these different businesses and their requirements will quickly improve and inform the capabilities of this system. Experience has shown that an open source approach will contribute to transparency, particularly in terms of underlying functionality and triage.
Open source is predestined to define standards that can be easily adapted across business boundaries. A powerful platform that incorporates all these elements and viewpoints is difficult to create and implement without a neutral community as the core development environment. The Enterprise Neurosystem open source community has already laid the groundwork.
And needless to say, Red Hat’s offerings such as OpenShift, Ceph and Middleware are a strong foundation for any Neurosystem deployment. These are platforms and tools that incorporate the same open source ethos.
An (enterprise) utopia
In summary, the potential advantages of the Enterprise Neurosystem are clear, optimised processes and deeper predictive capabilities, as well as greater accuracy, thanks to a core analytics engine that incorporates all AI models. This results in dramatic resource optimisations, and significant cost savings.
In particular, companies that have to deal with constantly changing circumstances are in line to benefit from obtaining deeper and more meaningful data sets. In the end-state scenario, this data will be processed by this primary artificial intelligence, and presented to relevant stakeholders in the form of observations and recommendations.
Importantly, the new capabilities free up people to focus on areas of innovation rather than mundane infrastructure operations, as machine support helps them achieve greater resilience and faster resolution timelines when problems arise.
Again, a major consideration for the success of the Enterprise Neurosystem project is ethics, and mitigating bias. This will continue to be an ongoing focus for the community, to ensure that AI evolves to improve societies and lives.
Ultimately, the Enterprise Neurosystem’s large scale interpretive framework will grow and evolve. It can eventually be used in a new manner an operating system for the planetary ecosphere. It could cross-correlate environmental data, assist in resource and species preservation, load-balance various ecological impact events caused by humans, and help mitigate the effects of climate change.
We’d like to offer an open invitation. If you’d like to join us and help define the future of AI in telecommunications, please feel free to reach out to us, and we’ll look forward to working with you on it.
The author is Bill Wright, head of AI/ML and intelligent edge, global industries and accounts, Red Hat.
Comment on this article below or via Twitter: @VanillaPlus OR @jcvplus
