AI is typically only available to a select few. In the majority of cases, data scientists with specialized skills would build algorithms and deploy models, while most others in an organization would simply see a black box.
But now technological advancements are empowering business users to take part in AI initiatives too — enabling them to easily build AI applications, access predictive insights and, most importantly, use these insights to make critical decisions.
I sat down with Ritu Jyoti, Program Vice President of AI Research at IDC to discuss this democratization of AI. We explored how making AI more accessible to people with a variety of skills can help accelerate time to value for businesses, and we talked about why she believes Palantir for IBM Cloud Pak for Data “creates magic in bringing teams as well as data and AI capabilities together in one cohesive fashion.” Now generally available, this powerful solution helps lines of business use predictive, data-driven insights to improve decisions and operations.
IBM: Let’s start with a staggering finding from your IDC AI StrategiesView 2020 research: Eighty-two percent of businesses are exploring or implementing AI. What is driving this surge?
Ritu Jyoti: AI has become much more accessible and commercially available in the last couple of years. People are realizing that when they base decisions on data and AI, they can speed time to value and enhance agility in real time. They are using AI to drive critical business needs like improving customer experience, boosting employee engagement, and accelerating innovation. That’s because advancements in machine learning and deep learning algorithms have reached a point where they can help make meaningful decisions. In addition, advances in computational power and cloud-native architectures have propelled AI adoption as well.
The pandemic has also contributed to the growing interest in AI, I imagine?
Yes, AI has played a key role in business continuity and resiliency. Conversational AI has helped contact centers provide 24/7, personalized customer service. AI has come in handy in making remote learning possible. And as we come back to the office, computer vision technologies can help enforce social distancing.
But AI implementation isn’t always easy. In the AI StrategiesView research, 34% of companies cite limited expertise as a barrier. Why is that?
There are not enough data scientists. There is a gap between the skills needed in organizations today and the skills acquired in schools. Reskilling people is crucial, and it’s extremely hard to implement AI by yourself. Many companies face a “build or buy” decision and try to “build,” but more than 50% of the time is spent on data preparation and data engineering rather than on actual algorithms and feature engineering.
How can democratizing AI address this skills gap?
Democratization makes AI available and accessible to the breadth of talent in an enterprise. Business users know the business in and out. Enabling them to build AI-powered applications using visual application development platforms, including those with drag-and-drop functionalities, can close the gap in data science talent.
It’s important to note, though, that AI democratization doesn’t replace data scientists. It helps business users collaborate better with data scientists and see them as a partner. Essentially democratization brings the business closer to the technology and the technology closer to the business. It can also free data scientists to do higher value work.
Most people think AI is all about technology, but it’s really more about business outcomes. Algorithms are important, but aligning to business goals brings greater relevance and competitive edge.
AI should become more inclusive with better synergies: Data scientists and AI engineers should build models alongside business domain experts —testing hypotheses, evaluating models, and repeating the process until results are acceptable.
Also referenced in the research was that “trustworthy AI is fast becoming a business imperative.” Can you talk about how AI democratization can support AI trust and transparency?
In the research, lack of trustworthy AI was in the top three challenges cited when implementing AI. When you democratize AI to include the broader workforce, it helps build up their trust. If they can experience AI first-hand and are part of an AI lifecycle that is transparent and trusted, they are more likely to embrace it.
Which milestones do companies need to achieve to democratize and scale AI?
At IDC, we have identified five dimensions – vision, people, process, technology, and data readiness – that can be used to evaluate an organization’s maturity for AI. “Thrivers” are those that embrace AI initiatives across all five dimensions to help drive digital transformation, achieve better business outcomes and sustain competitive differentiation.
How does Palantir for IBM Cloud Pak for Data democratize AI?
Industry use cases are going to be the disruptive force in AI adoption, and many of these customers are looking to simplify the end-to-end AI lifecycle.
Palantir for IBM Cloud Pak for Data creates that magic by bringing teams as well as data and AI capabilities together in one cohesive fashion: Data virtualization, data cataloging, machine learning, and explainable AI from IBM are paired together with Palantir’s business domain ontologies, low-code/no-code environment and templated solutions. It’s a one-stop-shop powered by Red Hat OpenShift, so you can run workloads anywhere and gain hybrid cloud flexibility.
Any last thoughts?
For businesses to stay relevant and competitive, AI is no longer a “nice-to-have.” Businesses worldwide are looking for technology suppliers to accelerate innovation and realize AI at scale. However, no technology provider can support all customer requirements alone. Leveraging ecosystems and partnerships is very crucial. Palantir for IBM Cloud Pak for Data is a classic example of a partnership that brings the best of each company’s AI and machine learning capabilities to the table.