Unleashing Business Potential with Gen AI

A guide for using Generative AI to transform Operations, Innovation, and Competitive Strategy for businesses and individuals.

By: John Kahan and Chad Richeson with the support of the Microsoft Word AI Copilot.

Preface

Over the past year we have seen an explosion in Generative AI (Gen AI) technology and solutions in the market. Individuals and organizations are rapidly working to take advantage of what is a once-in-a-generation technology wave that will transform work and businesses processes in ways not seen since the advent of the internet.  Because of our roles building data and AI solutions over the roughly past four decades – John as the Chief Data Officer of Microsoft, and Chad as Director of Big Data at Microsoft and at various startups -- many have reached out to us for help. While we can offer informed opinions, the truth is nobody has all the answers when it comes to AI. With that in mind, this article will offer our thoughts and insights from learnings derived over the past year plus dealing directly with Generative AI and the decades leading up to it. Comments and feedback are encouraged & appreciated.

Introduction

Artificial intelligence (AI) is rapidly transforming the business landscape, offering both vast potential and challenges. Among its forms, Artificial General Intelligence (AGI) and Generative AI (Gen AI) stand out for their capabilities and applications. AGI, still a developing concept, promises to perform a wide range of tasks with human-like understanding and adaptability, making it a key target for advancements in business, medicine, education, energy, and other fields. Meanwhile, Gen AI, a more specialized form of AI, is quickly gaining traction due to its ability to mimic certain aspects of human creativity and intelligence and solve problems beyond its initial training areas. These abilities enable Gen AI to perform complex tasks less expensively and more quickly than humans, which will make it not just a useful tool for businesses over the next few years, but an indispensable one.

Gen AI, however, is not without its complexities, notably its probabilistic nature, which means it can produce different outputs from the same inputs. This lack of full predictability can pose new challenges for certain tasks, but is also the source of Gen AI’s creativity. To meet this challenge, businesses need to learn new skills such as managing Gen AI with guardrails, and building new types of oversight mechanisms (that can also leverage AI.) That said, the true competitive edge lies not in the technology itself but in how businesses leverage unique data, manage and activate this data, and integrate AI seamlessly into operations. To attain this edge, Gen AI demands quick adaptation and experimentation from companies. Those that can effectively integrate Gen AI, paying close attention to data availability, model implementation, and ethical considerations, stand to gain a significant advantage in the marketplace.

The Unparalleled Potential of Gen AI for Business

Gen AI offers businesses adaptability, scalability, and enhanced learning capabilities that will drive enhanced productivity and speed, making it a transformative tool for businesses. It also will make your employees more productive and simply happier as they remove the mundane repetitive tasks from their roles and climb the important value chain of critical thinking and creativity. Unlike previous labor innovations like outsourcing, which offered benefits but lacked dynamic adaptability, Gen AI continuously learns and improves, handling a wide range of tasks efficiently and getting better as it goes. Businesses are starting to leverage Gen AI for a variety of applications, including enhancing customer service, personalizing marketing efforts, writing and debugging software, supporting product development, and refining decision-making processes. This marks a substantial advancement in how businesses harness technology for growth and efficiency.

The implementation of Gen AI, however, is not just about deploying a new technology -- it's about strategically integrating this technology into business processes, leveraging the company’s unique data and knowledge, and delivering the resulting capabilities seamlessly into customer and employee experiences. This integration is where the true competitive advantage lies, enabling businesses to not just automate tasks but to drive better outcomes, unlocking levels of efficiency and innovation that were previously out of reach.

Navigating the Rapid Pace of Gen AI Evolution

To embrace the opportunities afforded by Gen AI, agility is key. The ability to quickly experiment, learn, and adapt Gen AI solutions can give a business a substantial edge over competitors. The journey towards effectively adopting Gen AI requires consideration of several factors, including its impact on business operations, data availability and quality, the readiness of technical infrastructure, and implications for ethics, privacy, and security.

To harness the full potential of Gen AI, businesses must undertake a comprehensive evaluation across various domains:

1. Business Impact: Assessing the potential benefits and risks associated with Gen AI adoption is crucial. This includes examining how Gen AI can improve customer satisfaction, drive revenue growth, reduce operational costs, foster innovation, and deliver competitive advantage. Equally important is ensuring that Gen AI solutions align with the company's strategic vision, values, and ethical standards.

2.  Data Availability: The effectiveness of Gen AI hinges on the availability and quality of data. Adequate, diverse, and reliable data is essential for training, testing, and deploying Gen AI solutions. Companies must evaluate their data assets to ensure they can support the scope and complexity of Gen AI applications.

3.  Technical Readiness: The deployment of Gen AI solutions requires the right mix of LLM models, tools, and infrastructure. Companies must assess their access to and compatibility with LLM models, the ease of Gen AI integration into existing systems, and the scalability and robustness of the solutions. Monitoring and evaluation mechanisms are also vital to ensure the performance and adaptability of Gen AI applications.

4.  Ethical and Legal Implications: Gen AI's impact also extends to ethical and societal considerations. Businesses must navigate the ethical principles guiding Gen AI use, ensuring solutions adhere to fairness, transparency, and privacy standards. Legal compliance with data protection and intellectual property laws is also paramount.

5.  Security and Privacy: The adoption of Gen AI can introduce new security and privacy risks. Establishing robust security and privacy practices is essential to protect sensitive data and maintain trust. Businesses must address potential vulnerabilities, ensuring Gen AI solutions comply with data encryption, access control, and incident response policies.

Exploring Gen AI Applications Across Functions

Gen AI's versatility allows for its application across virtually every function, each offering distinct benefits and challenges:

• Marketing: Gen AI can revolutionize marketing strategies through content creation, customer segmentation, and campaign optimization, enabling personalized customer experiences at scale.
• Finance: From financial forecasting to fraud detection, Gen AI offers sophisticated tools for analyzing data, identifying trends, and making informed decisions.
• Operations: In operations, Gen AI can enhance efficiency through improved inventory management, demand forecasting, and logistics optimization.
• Customer Service: Generative AI enables 24/7 multi-lingual support that provides personalized, immediate responses to customer inquiries. It can also analyze customer feedback in real-time, allowing businesses to improve the customer experience.
• R&D: Gen AI can aid R&D personnel by generating hypotheses, designing experiments, generating code & fixing bugs, and creating & running simulations, which reduces time to market and increases quality.

Compelling use cases also exist for BI, Supply Chain, Human Resources, Legal, Strategy, and virtually every other enterprise function.

Gen AI Solutions

When developing Gen AI strategies, businesses currently have four primary solution types to consider, each with unique benefits and limitations and all keeping humans in the loop both expanding the capabilities of businesses and improving productivity and quality:

1.  Free Gen AI Tools: Platforms like Bing with the Chat GPT 4.0 Copilot, and Open AI offer free access to Generative AI, facilitating experimentation and work across various tasks. These free copilots enable anyone to use Gen AI to reason over, research, and create content using primarily web data.  They are a massive step forward from what we know of Search today.  In many ways, we are relearning to be human beings here moving from a world that Google taught us with keywords to a world where humans can interact as humans asking full scale questions and continuing a conversation until the desired content is formulated.  While anyone can access these services it is important to understand the results need to be checked.  Gen AI is just a statistical algorithm that guesses at the next word or character in a sentence.  It occasionally gets things wrong and thus checking citations and/or asking for them is required not unlike Search today.  In addition, while there have been published reports of leaking data, this only occurs if you publish your data on the open web.  These models are trained offline and not in context to your searching, however, many of these services have terms and conditions that align to the advertising models of Search. E.g., they can send you targeted advertising based on your conversations across services. You do have control over this in your browser settings.

2.  Paid Copilots: Subscription services like Microsoft M365 Copilot provide more advanced features, including better integration, data analytics, and support, customization, and security. These are an excellent way to get started as they enable you to reason and create not just on web data but on your own individual data.  They also offer terms that specifically guarantee they will not use your data in any scenario beyond your session. Early results indicate that productivity gains here are very high in both task completion and in quality of work, especially for newer employees. One large recent study released by the National Bureau of Economic Research found that customer support agents who had help from a chatbot were 14% more productive, on average, based on the number of issues they resolved per hour. The AI-supported agents ended conversations faster, handled more chats per hour, and were slightly more successful in resolving problems. Notably, the effect was largest for the least skilled and least experienced workers, who saw productivity gains of up to 35%. Another study performed by MIT and Harvard found that Boston Consulting Group employees achieved 40% higher productivity with an AI assistant versus a control group who had no access to AI. These gains are unprecedented according to Erik Brynjolfsson, a Stanford professor who studies the impact of information technology on businesses, who stated that, “often companies are happy to get 1% or 2% productivity gains.”

3.  Tailored LMM models and solutions: Solutions designed for specific domains, like customer service or marketing, offer more control and customization. Businesses can leverage the large models created primarily by tech companies such as OpenAI, Anthropic and others, tailoring them with their proprietary data and open data into solutions specifically designed for their business using techniques such as Fine Tuning or Retrieval Augmented Generation (RAG). The resulting solution delivers specialty capability and unique competitive advantage; as well, these solutions will likely require changes to business processes and the roles of individuals.

4. Custom Gen AI LMM models and solutions: Businesses with very unique needs can develop fully customized Gen AI models and solutions, including creation of their own foundation LMM models. Owning the entire Gen AI value chain offers the opportunity for unparalleled innovation and performance; however, this path requires strong data science/AI skills and large-scale compute resources.  This approach is currently only warranted for specific cases, such as for personalized medicine, financial fraud detection, autonomous vehicles, etc. Due to the high cost, only a fraction of businesses are expected to build and maintain their own foundation models.

Each option presents a trade-off between ease of use, customization, cost, and the level of innovation they enable, guiding businesses in aligning their Gen AI strategies with their overall goals and capabilities.  It is also important to understand there are many models out there today and many more are coming.  While we hear about Chat GPT there are several versions of this including Chat GPT 4.0, Chat GPT 4.0 Turbo, etc. and many others such as Claude, Gemini, LLaMa, and even small language models.  All are trying to deliver accurate results as possible, with speed and at the optimal cost model.  Not all solutions will require the top models to operate to get value. Businesses and individuals should learn work works best for each scenario and keep learning as new models and solutions are developed and delivered to the marketplace.

Conclusion

The strategic implementation of Gen AI presents a transformative opportunity for businesses and individuals willing to navigate its complexities. By carefully evaluating the potential impacts, readiness, and ethical considerations, companies can leverage Gen AI to drive innovation, efficiency, and competitiveness. The journey requires a commitment to continuous learning and adaptation, but for those who embark on it, the rewards can redefine the possibilities of their business operations and strategic outcomes.

John Kahan is the former Chief Data Officer at Microsoft and currently serves as a Board member and advisor to over a dozen companies globally.

Chad Richeson is the former Director of Big Data at Microsoft and recently co-founded Firebrand.ai, a startup that helps enterprises adopt Generative AI-driven workflows.