The Power of Generative AI for Business: Insights from IBM THINK 2023

IBM THINK 2023 showcased the incredible potential of generative artificial intelligence (AI) for businesses. From AI-generated art to AI-generated songs, the possibilities are endless. However, when it comes to applying foundation models in big business, a higher standard is required. AI should be trusted, secured, and adaptable.

Building Trustworthy and Scalable AI Solutions

Unlike simple automation that is trained for a specific task, AI in the business world needs to be versatile and integrated across various organizational functions. It should not be limited to a single system but rather designed as hybrid-ready AI that can scale across different systems. Transparency is crucial – businesses need to know where an answer or solution comes from and how it was derived.

By embracing AI as the core of their operations, businesses can unlock new opportunities and achieve greater success. This goes beyond just using AI; it’s about becoming an AI value creator.

Becoming an AI Value Creator

Instead of being merely an AI user who relies on someone else’s model without control over data or the model itself, businesses can become AI value creators. As an AI value creator, you have multiple entry points and avenues for innovation:

  1. Bringing Your Own Data: You can leverage your own data and integrate it into Watsonx, IBM’s integrated data and AI platform.
  2. Accessing a Library of Tools: Choose from a wide array of tools and technologies available in Watsonx for training, tuning, validating, and deploying both traditional machine learning models and foundation models with generative capabilities.
  3. Influencing Training: You have the ability to train and influence the training process of your models.
  4. Transparency and Control: With Watsonx.governance tools, you can ensure responsible execution of your AI solutions while maintaining transparency and control over both data governance and AI model governance.
  5. Iterative Improvement: Through a creative process, you can prompt, improve, and iterate upon a family of models instead of being limited to a single pre-existing model.
  6. Ownership of Foundation Models: Foundation models trained with your data become your most valuable assets. As an AI value creator, you own these models and all the value they generate for your business.

Introducing Watsonx: IBM’s Integrated Data and AI Platform

To empower businesses in becoming AI value creators, IBM has developed Watsonx – an integrated data and AI platform consisting of three primary components:

  1. A massive curated data repository that brings together raw data from various sources, including public sources and IBM proprietary data. Clients can also enrich their purpose-built foundation models by bringing their own data to this repository.
  2. An enterprise studio that facilitates training, validation, tuning, and deployment of traditional machine learning models as well as foundation models capable of generative capabilities.
  3. Watsonx.governance: A suite of powerful tools designed to ensure responsible execution of AI solutions.

These three components seamlessly work together throughout the entire lifecycle of foundation models. Built on Red Hat OpenShift, Watsonx also offers seamless integration with other systems and allows for deployment in any IT environment.

The AI Workflow with Watsonx

Deploying complex technologies like AI at the enterprise level can be challenging. However, Watsonx simplifies the process through its comprehensive platform. Let’s take a closer look at the end-to-end workflow within Watsonx:

Data Preparation

Data scientists can easily access data from different sources (such as public clouds or on-premises databases) through The platform establishes necessary connections between these sources for easy access. IBM has amassed vast amounts of raw data across multiple domains, including the internet, code repositories, academic sources, and enterprise data. Clients can also bring their own data to enrich the training process.

Data preparation involves categorizing, classifying, filtering, and processing the data. allows for seamless management of different versions of data through metadata that ensures traceable governance.

Model Training

With, data scientists can select a model architecture from IBM’s five model families. The chosen architecture serves as the foundation for training and fine-tuning the models using computing resources across hybrid cloud environments. Tokenization is performed to break down sentences into tokens, which are then used to train the model.

Training is a complex and time-consuming task that requires significant computational resources. leverages open-source technologies to simplify the training process and auto-scales resources accordingly.

Embracing AI as an Opportunity

IBM THINK 2023 reaffirmed that we are in the midst of a defining moment for foundation models and generative AI. To fully embrace this opportunity, businesses must become AI value creators rather than solely relying on pre-existing models.

By leveraging Watsonx, businesses can harness the power of generative AI with proper governance and control over their own destiny. The possibilities for innovation and value creation are immense across industries such as customer care, logistics, medicine, manufacturing, energy, automotive, aerospace, communications, and beyond.

It’s time to think bigger and create a future where AI is not just a tool but an integral part of business success. Let’s embark on this journey together with IBM’s Watsonx platform.

Disclaimer: This article summarizes key points from a session at IBM THINK 2023 featuring Dr. Dario Gil.

Leave a Comment