Artificial intelligence is a complicated matter, and a solid ecosystem is the foundation of successful development.
In creating an AI system, my team worked to determine what the “right” ecosystem for AI looks like. For our system to continuously become smarter, we knew an ecosystem made up of technology, data, industry-specific knowledge, research, and security would be necessary.
Without each of these key ingredients, even a human wouldn’t be able to execute what we ask of AI, meaning the ecosystem plays a vital role in such an initiative’s success.
Challenges in AI development
In order to determine the best environment for artificial intelligence, it’s important to first acknowledge the challenges AI faces.
Everyone is talking about AI, but that doesn’t mean it’s not a nascent concept. Many companies are still experimenting with it, so there’s no de facto standard for establishing the proper ecosystem. Companies have implemented systems and witnessed proof of their promise, of course, but it’s still too early to treat success stories as role models.
The technologies, too, are maturing, so systems’ nuances will vary across customers and processes, and form factors change by the quarter. It’s a trial-and-error model that has its challenges, so it requires practitioners to pioneer it and make judgment calls.
Like any software development endeavor, creating an AI initiative results in a lot of changed and abandoned courses with several disappointments. Building something out of nothing is rarely a simple task. Short-term issues (such as automation) can take away the bulk of the work while people get carried away, siphoning resources for projects that don’t support the long-term successes.
In overseeing our AI system’s development, I found that having the right ecosystem in place won’t remove these challenges, but it will help companies overcome them. Below are four key elements of an AI ecosystem that will set companies on the most straightforward path to success:
1. An innovation center: There’s a lot of hype around AI, and it becomes difficult for organizations to separate that hype from reality across various stakeholders. An innovation center acts as a specialty clinic, where all users can get their processes and an expert team that understands the technology and existing constraints.
In an innovation center, companies can develop pragmatic, implementable solutions for each process in an organized way. It encourages project success, standardization across the organization, and more optimization and reuse of the initiatives, reducing the risk that accompanies experimenting randomly and recklessly.
2. A central governance environment: AI’s fuzzy logic allows it to think, learn, and adapt on the fly, which, while impressive technologically, isn’t always good for business. Microsoft’s Twitter chatbot, Tay, showcases this. Behaviors of AI vary from instance to instance, and it’s important to put principles in place that control rogue behavior, especially in a system’s formative years. This is necessary to prevent risk and to build scalable systems.
3. A data curation service: All companies have data, but making it qualitatively rich and comprehensive isn’t an easy task. AI requires access to a digitally rich and diverse cache of organized data. Companies should enrich existing data sets or, if that’s not possible, buy external data to complete the scenarios. It’s an ongoing exercise that needs to be leveraged across the organization to be sustainably effective.
4. A holistic mindset: Ecosystems are inherently integrated, so those in charge of an AI system must keep a holistic mindset. To do so, companies should maintain a reference architecture that evaluates the system comprehensively — ideally, one that examines the processes on a 2D matrix of repetition versus cognition.
The framework can evaluate the process in terms of ROI, digital readiness, ease of implementation, customer/user impact, business criticality, compliance issues, and more for prioritization. It should be considered holistically, with both the front and back offices considering each other’s efficiency, effectiveness, and experience.
There’s no magic in AI; it takes time, effort, data, and expertise to build a meaningful system. Twenty years ago, it was only available to the largest and most advanced enterprises, but it’s quickly becoming a business standard. Like the Internet that fuels it, AI is inevitable. Having the right ecosystem in place will ensure companies are prepared when they meet this inevitability.
K.R. Sanjiv is the chief technology officer for Wipro, a global information technology, consulting, and outsourcing company. Sanjiv has more than 25 years of enterprise IT experience, including consulting, application development, and technology development spanning multiple industry segments and diverse technology areas. Sanjiv holds a bachelor’s degree from the Birla Institute of Technology and Science Pilani.