Generative AI has the power to ignite innovation across infrastructure, models, and apps — providing organizations with the ability to unlock the value of data and insights quickly, efficiently, and responsibly. It also will, undoubtedly, change our business and private lives.
Srijani Dey and Tab Chowdhury, who have been deeply immersed in emerging technologies throughout their careers, have seen mind-boggling change already. They had a lot to say about how generative AI is bringing unprecedented change, and offer a vision for what could happen next.
Generative AI is not new per se — but it’s suddenly a topic in every board room. What’s different now?
Srijani Dey: Generative AI has pushed the boundaries, and industry is both excited and apprehensive of the AI landscape now. There is a change in the mental model as to how leaders are envisioning the future. Previously, tech giants and large-scale industries were harnessing AI for profitability and competitive edge.
What organizations are going to see now is a focused effort to operationalize and industrialize AI across business sizes and industries. By simplifying implementation and reducing price points, AI can expand its reach — similar to the Law of Accelerating Returns.
Business is now pivoting from a “so what” question to a “why don’t we” mindset — while looking for measurable outcomes within compliance, ethical, and legal guardrails. As business leaders, it’s important to adopt the technology and collaborate to identify what problems need to be solved. But your mileage will vary on how ambitious you want to get in institutionalizing AI as part of your organizational culture.
Tab Chowdhury: Historically, a barrier to business leaders adopting generative AI was cost and complexity of the technology required to implement solutions. Business leaders are now noticing that this barrier is no longer an issue as organizations like Amazon Web Services (AWS) have built low-cost, purpose-built offerings. For example, AWS is offering tools that help democratize the development of generative AI solutions, enabling businesses to improve efficiency and reduce downtime and cost while delivering innovative solutions to its customers.
What might business leaders misunderstand about generative AI?
TC: Business leaders need to realize that generative AI is a tool and requires context and focus to solve problems. The keys for business leaders include identifying current problems that could be solved with generative AI and thinking about innovative solutions that could be disruptive.
SD: Removing the complexity of implementing this science will help adoption. Then the science becomes an art — for example, how swiftly you navigate through your enterprise boundaries, organizational seams, and lines of business for the adoption of the benefits of generative AI.
What will the AI ecosystem look like in the near future?
SD: In order to embed generative AI organically in enterprises, there should be a focus on value stream mapping — that is, providing context for how to apply the technology, and align it with the right set of legal, security, and ethical compliance to put guardrails around use. Enterprises will also need to drive a high degree of interoperability and transparency to maximize the reusability of generative AI applications.
We are in an era where there are parallel accelerating technologies like quantum, neural networks, generative AI, augmented reality and virtual reality (AR/VR), and nanotechnology. These are extremely powerful by themselves and are driving tremendous momentum. But siloed models do not work. Our key question is: how do we bring these developments and people from multidisciplinary teams together to create potential innovation? If these technologies can be leveraged together, in a multimodal fashion, this will be a revolutionary breakthrough.
For example: at DXC, we have been able to detect scratches on a car in a manufacturing line using computer vision models — and, as a result, reduce the return from dealers. However, we could not get beyond 84 percent accuracy as there were not many images for edge cases (such as a scratch on a curved or reflective surface). Through multimodal generative AI, we generated new images and retrained the computer vision model, which increased the efficacy of the existing computer-vision model by three percent. Read more about this use case
What are your recommendations for adopting generative AI — the right way?
SD: Just because generative AI is here, don’t try to force-fit it. It’s not just the technology that should lead. The value realization happens when you have an alignment of why, what, and how. In fact, surveys have revealed a hindrance for the adoption of AI-driven decision intelligence: 8 percent of the impediment is due to technology, and 92 percent is due to people, process, and culture.1
This can be addressed by an impactful AI strategy, clear operating models, and crisp value realization definition. At DXC and AWS, we can be the enablers, but you also have to start thinking outside the box about how and where your organization can deliver value. As a decisionmaker and leader, you know your business processes — you know what needs to be done. A few questions can help you get started: Where exactly do you want to integrate? What process inefficiency are you trying to solve? Success will occur when we are able to harness the collaborative efficiency of biological and machine brain’s “day in the life” augmenting each other — and that becomes the new normal. Success will occur when we are able to harness the collaborative efficiency of biological and machine brains to augment one another — and that becomes the new normal.
TC: Some businesses are well advanced and have all the necessary pieces in place; some are in their infancy. The recommendation is to start small with a proof of concept (POC) using a non-critical workload, measure success, and then go big. AWS offers business leaders work-backward sessions focused on outcomes. Together, we create an action-driven plan to get there. AWS enables customers in each step of their generative AI journey with tools, training, skilled resources and innovation funding to ensure a successful outcome.
As generative AI evolves, what could be the next revolutionary breakthrough — and how will organizations react?
SD: Whenever there is a technology shift, there is apprehension. Think about the evolution of cloud technology and infrastructure — it had the same bell curve of apprehension. Yet cloud modernizations are table stakes now. We are moving to the next innovation surge with generative AI. In order for generative AI to get absorbed into culture organically, we need to let it go through the four stages of human habits (credit to James Clear’s Atomic Habits):
- Noticing: Acknowledging that change is imminent.
- Wanting: Training the workforce to be ready for the revolution.
- Doing: Starting the adoption journey, and implementing and scaling use cases.
- Liking: Commoditizing and industrializing.
It is important for every organization to assess which stage of generative AI evolution they are currently in to plan the strategic trajectory of adoption.
There are organizational changes that are imperative and set directives can guide you. For example, you want to set an objective metric with stakeholders and then work backward to achieve it. Your measurable qualitative and quantifiable outcomes need to be aligned to and reflective of organizational benefits. Nearly every organization has taken some form of AI initiative; however, the bulk of the AI models don’t go to production — 87 percent of data science projects never make it into production, for example.2 To make it effective and impactful, think about what you can do differently with generative AI. And, to see a vision-to-value conversion, you need a strong collaboration between business and IT, and you need an authoritative decisionmaker to move fast — else the competitive edge is gone.
TC: Anytime businesses are trying to do something different, innovative and new, there will always be resistance. Disruptive technology brings opportunities for innovation, and it starts with adoption of an organizational change management (OCM) plan that includes a team operating model, clear deliverables, a training plan, and measurable success criteria. There will be new training, new tools, new processes, and a new way to work together. Businesses need to ensure they do not get caught up in a turf war but rather focus on the outcome with an appropriate operational model in place that will remove any friction among teams.
What should business leaders know about working with DXC Technology and AWS to adopt generative AI?
TC: Software developers today spend a significant amount of their time writing code that is pretty straightforward and undifferentiated. They also spend a lot of time trying to keep up with ever-changing tools. All of this leaves developers less time to develop new, innovative capabilities and services.
AWS offers customers tools — such as Amazon Bedrock, Amazon CodeWhisperer, Amazon SageMaker JumpStart, and Amazon Titan foundation models—and provides the flexibility to work with open-source models or build their own foundation models. AWS has also built vector database capabilities in Amazon OpenSearch, Amazon Aurora, Amazon Relational Database Service, and purpose-built chips AWS Trainium and AWS Inferentia2 for generative AI. With AWS, all of this comes together — the data, the tools, the business solutions. Everything is connected to provide holistic value.
Read more about DXC generative AI
SD: If you need to embed generative AI models at the application, you will need integration, middleware, and frontend application skills, but you may not have these on your team — and it’s a herculean task to bring together a multidisciplinary team. AWS focuses on solutions and endto- end implementations. Having an end-to-end solution enables faster time to market. AWS is targeting to address the right set of problems around massive scaling of the solutions. For example, AWS offering purpose-built AWS Inferentia hardware chips really helps address solution scaling challenges.
DXC Technology brings unique accelerators to strengthen your foundational capabilities — like data quality, metadatadriven design, and transparency through lineage. These are a critical part of the solutions that you need to have to make AI a reality across the organization. There are also accelerators for standing up generative AI services with required guardrails for drift detection. Additionally, DXC has a decision intelligence and AI advisory committee that brings together the process of how to operationalize AI and drive value realization across an organization.
The most performant and cost-effective tools and infrastructure for generative AI.
AWS has been training for this moment. Generative AI isn’t new, but deep expertise refined over years of experimentation at AWS is newly available to your business in a range of easy to adopt, low-code tools to help you:
- Innovate: Create generative AI applications that unlock the value of your data quickly, captivate your customers with new experiences, and potentially generate new revenue streams.
- Scale: Customize models with enterprise data, avoid third-party lock-in, and confidently scale generative AI, knowing your data will not be used to inform models.
- Perform, responsibly: Get support filtering model outputs for unacceptable content.
You can even save on energy costs and work toward sustainability goals because AWS silicon chips in Amazon EC2 instances for machine learning are designed to be energy efficient.
With AWS, you don’t need to invent nor reinvent generative AI capabilities — they are ready to deploy now. You just have to decide what business capabilities to reimagine first.
Reduce costs by 50% when you train models on Amazon EC2 instances powered by AWS chips purpose-built for ML.
Start here:
Any generative AI project can benefit from these powerhouse tools.
Amazon Bedrock
The easiest way to build and scale generative AI applications with foundation models.
Amazon CodeWhisperer
Save developers time and get applications to market faster.
Amazon SageMaker JumpStart
Deploy prebuilt machine learning solutions in just a few clicks.
Where do you see generative AI going next?
SD: It is important to note that AI will not replace humanity. Rather, humans with AI will outperform humans who do not use AI. It’s possible that we can even be looking at a productivity level to enable us to enjoy a three-day weekend with our friends and family.
TC: The sky is the limit. As Srijani mentioned, generative AI will not replace humans but rather augment what we do every day to deliver business outcomes in a more efficient way. Think of it as a collection of assistive solutions and tools that could be utilized to make things cheaper, faster, and better. At AWS, we are continuously developing new foundation models, tools, industry-focused solutions, and offerings to solve every day problems and make generative AI adoption easy for our customers.
SD: Yes, there is excitement and anxiety, there is speculation about how to drive economic and even social progress with AI. Maybe kids going to middle school will have AI in their curriculum. Life will never be the same again. But this is a move in the right direction, and this wave has started making people think harder.
1 Sloan Management Review, “Execs Bullish on AI but Wary of Data Leadership,” March 2021
2 VentureBeat, “Why do 87% of data science projects never make it into production?” July 2019