Read the original article in Forbes here.
As AI continues to touch all areas of an organization, from customer experience to operations, CIOs are faced with taking a hard look at their technology infrastructures to see how they're using AI and what the best use of it is. In many cases, CIOs must play the role of chief AI officer, deliberating over which vendors to work with and how to govern that approach throughout the organization.
Understandably, CIOs are under increasing pressure to grasp all aspects of AI while seeking total visibility into how departments leverage the technology. Because of this responsibility, a CIO might hitch their cart to one vendor to ease their AI strategy and governance. However, others will argue the approach can limit a company's success and even put them at risk.
Weighing A Single-Vendor Vs. Agnostic Approach To AI
On the one hand, locking an AI business strategy to one vendor helps standardize how a company uses the technology. For example, a company might use only Microsoft or Google and then build solutions on top of these cloud provider options.
On the other hand, locking into one provider can limit long-term success and stifle innovation. For this reason, companies in any sector need to know they can still govern, standardize and tap into multiple sources of AI.
By steering away from the convenience of one single provider, a company can be AI-agnostic and embrace a wide range of providers. Another way to think of this is like diversifying investments in a stock portfolio. Companies can mitigate business risk, highlight stable AI performers and accelerate how they innovate.
Reviewing AI Providers And Technologies In Many Shapes And Sizes
When CIOs begin to study different AI vendors, whether to use multiple solutions or lock into one larger provider, some important elements to consider include:
Infrastructure, Scalability And Cost
Major cloud providers like Google Cloud Platform (GCP), Amazon Web Services (AWS) and Microsoft Azure offer highly scalable infrastructures. Organizations want to review how the vendors will work within their infrastructures in terms of service speed but also allow a business to tune for cost to maintain tight budget controls.
Model Capabilities
Different AI providers offer models with varying strengths and capabilities. Organizations gauging multiple vendors will want to see how some models excel at natural language processing, while others are better suited for computer vision or code generation. It's also about finding models that fit different teams within an organization.
Specialization
Many AI providers specialize in certain markets, like healthcare-focused vendors with tools and models that fit the language and ways healthcare companies work. These specialized vendors may have use cases that give businesses new reach with fast ROI. At the same time, hyper-domain-specific models and analytics that are both commercial and open source can tackle tough problems, but they require expert oversight.
Integration And Tools
The level of integration with existing systems, workflows and development tools can differ significantly. The "how" is just as important as the "what" when it comes to putting AI technology into production. CIOs should ask: Does the AI connect via APIs? Can it read my company's data? Does it have a UI or connect to our company's IT ecosystem?
As CIOs evaluate AI vendors, they should review how models are specialized for a particular market, how the tools integrate and the strengths and weaknesses of different models to find the right fit for their company. Being agnostic to the provider itself and solely focusing on the value the solution brings to a business can also help the AI thrive.
Implementing AI Solutions Requires Discipline
In the rapidly evolving landscape of AI, CIOs come across technologies and platforms at a dizzying pace. Companies need to be both flexible and stringent in how they tackle AI, following a disciplined approach to how they review vendors. This means consistently evaluating new solutions against established criteria, loosely coupled to specific platforms and resisting the temptation to become overly reliant on any single AI ecosystem.
For all AI platforms, technologies, projects and use cases, collecting, staging, automating and tracing the data being fed into a company's applications is as important as the model itself. Avoid investing time in unifying data and intelligence into proprietary or closed platforms. Instead, CIOs need to seek the best method and tools for their team to centralize and share AI data across the enterprise as well as the full AI portfolio.
For these reasons, an agnostic view of AI keeps a company's focus on the business.