This talk was delivered as a keynote at Google's Pittsburgh office for a Pittsburgh Business Times event on December 13, 2023. It was never recorded. Two years later, the core thesis — that people underestimate AI because they underestimate language — has only gotten more true. In January 2025, the World Economic Forum updated its projections: 170 million new jobs created and 92 million displaced by 2030, with 86% of businesses saying AI will transform them. The numbers got bigger. The framework didn't change. Here is the talk as it was written.
Good morning,
Thank you for having me. It's a thrill to be standing here in this space. It was Google of course who introduced the world to the attention mechanism and transformer architecture that has led to these modern AI systems we'll be talking about today.
I was asked to talk to you about the future of AI in the workforce.
I have some nice stats and BCG research that I'll be sharing about that, and I'll also be sharing some of my own thoughts about where this is heading.
But before we discuss the future, let's take a step back.
Way back. To 335 BC, the Lyceum in Greece. Aristotle believed deeply in the power of language.
He encouraged his students to think carefully about how they used words to describe, to persuade, to reason.
Language does more than facilitate communication; our reality is shaped by the language we use to describe it.
The introduction of new terms and concepts can lead to profound shifts in societal norms and scientific understanding.
The transformative power of language is evident in every major shift in human society. The Renaissance, the Enlightenment, the Industrial Revolution — each era brought new terminologies, new narratives, and with them, new ways of understanding and interacting with the world.
The evolution of language is not just a record of human progress; it's the driver of it. It's a powerful, multi-purpose adaptation.
The story of AI as it exists today, is in a sense, the most recent chapter in the story of the evolution of language.
The models used by AI systems like ChatGPT and Gemini are known as large language models. Large language models are essentially statistical representations of language. AI systems that are built on LLMs are, in a sense, language engines. They generate language. And I believe that just as steam engines harnessed physical energy to power the Industrial Revolution, language engines will power a new kind of revolution. One that will reshape not only our workforce, but will amplify our ability to leverage our most powerful tool, our language, to reshape our culture.
Augmented Intelligence
Let's discuss the language we use when we talk about "AI." The term 'Artificial Intelligence' carries certain connotations and cultural biases. It suggests a form of intelligence that is separate and distinct from our own — something not quite real.
But what if we were to take a cue from Aristotle, and consider our use of language here?
By shifting our terminology to 'Augmented Intelligence,' we reframe our perception of AI's role in society to align with our current understanding of AI as a technology deeply integrated with human input and human interaction.
We provide ourselves with the language to leverage AI as a collaborative force.
The Power of Conversations
I like to think about a conversation as a game of language.
It's a game we play with words, taking turns, working together, trying out new patterns of language in an attempt to describe the future we'd like to build.
The game of language is an open-ended one, an infinite game in which each and every one of us is engaged throughout our lives. With each conversation we push our collective understanding. We gain leverage.
My introduction to AI began with a conversation. Two conversations, in fact.
For the past five years, I've been part of BCG, where my role has involved translating complex technology concepts into engaging narratives for a business audience.
In this role, I've interviewed hundreds of technologists and designers, people at the forefront of building the future.
One of these people was Kent Vasco, a colleague at BCG known for his expertise in creative technology and engineering.
Four years ago, I interviewed Kent as part of a video series and at the end of our interview, I asked him about what he saw as the future of technology. His response was simple: 'Have you heard of machine learning models?' Although I had some awareness, this question prompted me to delve deeper. I began researching neural networks, GANs, and the innovations being pioneered by teams like Google's DeepMind.
Several months after my conversation with Kent, I had the privilege of interviewing Refik Anadol, an AI artist renowned for his groundbreaking integration of technology and art.
Refik introduced me to the concept of 'latent space' — the unseen, multidimensional space where AI algorithms use relationships from their training data to generate new combinations of data.
As I began to work with these early models, exploring the potential of this latent space, I began to see AI not just as a tool but as a medium in its own right, and a partner in the creative process.
It was the beginning of a conversation with machine intelligence.
Conversations create new language. New language creates a new reality.
Every new term, every novel syntactical construction, expands our ability to think, to invent, to describe, and thus to create the world around us.
We create the future first by describing it.
In 2019, I began to experiment with GPT-2, an early precursor to the models used in ChatGPT, and I was astonished by its ability to reason and generate predictive text.
I learned to fine-tune the model on my own datasets and started using it in creative projects.
In 2021, I discovered GPT-3 and was blown away. It felt like interacting with a mind that could understand and expand upon ideas in a conversation.
Then, in November of 2022, OpenAI released ChatGPT. This moment was a turning point in the perception of AI within the business community.
As generative AI gained traction, my focus shifted towards its practical applications, working with BCG teams and clients, coaching them on AI best practices for prompting and effective AI system design.
In early 2023 a client asked us to envision the future of hyper-personalized advertising through generative AI.
Teaming up with Kent Vasco, the engineer who had first introduced me to AI, we worked to design and build a system that used a chain of AI agents, large language models, image generation models and speech synthesis models designed to showcase 'the art of the possible' in AI's capability to transform traditional advertising methods.
So that's a bit of background. Now let's take a look at the state of AI in business today.
In the past year, I've noticed a huge disparity in the understanding and adoption of AI across society, and especially in business.
"The future is already here — it's just not evenly distributed." — William Gibson
When it comes to AI adoption and the competitive gains that are realized, the future is already here. It's just not evenly distributed yet.
We are now witnessing the emergence of a new "digital divide" in business.
The Digital Divide
Let's talk about the four categories of AI adoption that have emerged over the last year, drawing on insights from BCG research:
The Avoiders: Over half of executives are hesitant to integrate AI into business operations, primarily due to concerns about practicality and a lack of understanding of potential applications. As AI advances at an accelerated rate, these companies risk quickly falling behind competitors and upstarts who have embraced the gains of AI adoption.
The Dabblers: These teams are experimenting with AI, but have yet to put any policies in place and have not yet done any work in preparing their workforce for the arrival of AI. According to the BCG Digital Acceleration Index, 37% of organizations now fall into this category.
The Adopters: A growing number of organizations are actively integrating AI to enhance operations and decision-making. These organizations view AI as a source of value going forward, and are beginning to setup guardrails and policies to support teams. While optimistic about AI's ROI potential, 80% face challenges such as the lack of a strategic roadmap for AI integration, indicating complexities in adopting and adapting to AI technologies.
The Innovators: AI innovators are not just using but also shaping AI's future in the workforce. Think OpenAI, Google, Microsoft, Nvidia, as well as the big three consultancies. Characterized by heavy investment in R&D and a more prevalent daily use of AI within the workforce, they navigate ethical standards and regulatory landscapes. This pioneering approach positions them as industry leaders in AI utilization. Only 3% of companies currently fall into this category.
Think about where your organization exists in this adoption curve. Think about where you yourself fall as an individual.
Now, we can get a bit of a preview of what the future of AI in the workforce will look like by studying the impact of AI on these early adopters.
The Leveling Effect
In a Boston Consulting Group study involving 750 consultants, AI's role in driving value across sectors was examined, revealing a few key insights.
One of the findings was the leveling effect of generative AI: when used for creative ideation, nearly all participants, regardless of their initial proficiency, showed improved performance. This study suggested that the use of AI tools can democratize performance in creative tasks, effectively narrowing the gap between varying skill levels.
Value Creation vs. Value Destruction
In the same BCG study, using AI for creative ideation tasks resulted in a remarkable 40% improvement in performance compared to those working without it.
However, the same study also indicated a 23% decrease in performance when AI was applied inappropriately. Challenges such as AI 'hallucinations' producing misleading outputs and a lack of diversity in AI-generated ideas underscore the need for careful, strategic deployment. This involves leveraging AI's strengths while mitigating its weaknesses through responsible and informed use.
The state of AI in business today is marked by its immense potential for innovation and efficiency. However, it's tempered by the need for careful integration, ethical consideration, and management.
Navigating this dual narrative of AI involves striking the right balance between leveraging AI for its strengths and maintaining the essential human touch in business operations.
In a recent article published in Fortune, BCG's Francois Candelon compares the current technological upheaval to the onset of the industrial revolution, signaling an enduring and accelerated transformation in both business practices and societal functions.
In response to this imminent change, Candelon suggests that businesses must develop robust transformation plans, implementing two-way information flows in technology architecture and engaging in strategic workforce planning.
The World Economic Forum predicts a significant shift in the job market due to AI: a 40% increase in AI and machine learning specialists by 2027, and while 85 million jobs may be replaced by 2025, an even greater number, 97 million new jobs, are expected to be created.
Over 120 million workers globally are projected to require retraining in the next three years to keep pace with AI's influence.
These startling predictions point to a pivotal shift in the global economy and workforce, demanding an urgent response from business leaders: both businesses and individuals must adapt to a landscape where AI skills are crucial. It underscores the need for extensive retraining and reskilling, preparing for a future where AI is integral to every aspect of work and economic contribution.
The 10/20/70 Principle
This permanent AI revolution is not a fleeting trend but a new reality. To navigate this landscape, BCG proposes a practical and strategic framework: the 10/20/70 principle, a guideline for effectively adopting AI in a business context.
According to this principle, the initial 10% of the effort in AI adoption is focused on developing robust AI systems.
The next 20% of the effort goes into ensuring quality data and innovation in technology implementation.
But the most significant effort involved in AI adoption in businesses is not about the technology.
It's about people.
70% of the effort in adoption is dedicated to transforming business processes and functions. This involves rethinking and redesigning how business operations are conducted in light of AI capabilities — aligning business goals, creating new workflows, and ensuring that the organization as a whole adapts to the AI-driven changes.
Change Begins with a Conversation
Moving forward, the most urgent advice I would give to anyone aiming to grasp AI's impact is to gain real-world, practical AI experience.
Merely reading articles and attending conferences is insufficient.
Brief interactions with free, web-based tools using older models won't cut it.
To truly understand AI, business leaders must immerse themselves directly in AI systems. This means working hands-on within AI playground environments, studying how changes in prompts and parameters influence outcomes.
It involves learning how to integrate AI into applications and operations and encouraging teams to do the same.
Perhaps most crucially, it requires continuously engaging in ongoing conversations with the latest AI models.
Yes, you heard me right. I'm literally asking you to talk to a machine.
Every day.
You see, AI skills differ fundamentally from traditional technical skills. They are rooted in the mastery of language — the ability to accurately position a problem with appropriate context, provide a clear vision, and collaboratively guide the AI toward a solution through iterative adjustments via natural language conversations.
By engaging in conversations with these AI systems, we gain a broader perspective. We gain an ally in the game of language.
The integration of AI into society and business is not just a technical challenge; it's a human challenge.
Embracing this challenge begins by shifting our focus from the technology of the moment, which is ever changing, and focusing on our use of language as the underlying essential tool in our toolkit.
Just as the invention of the printing press expanded the reach and impact of the written word, AI systems are expanding the potential of human language, enabling problem-solving and transformation at unprecedented scales — fostering new forms of expression, enhancing human creativity, understanding, and connectivity.
Thank you.
Bill Moore is a creative director and systems designer based in Pittsburgh. He has spent the past eight years at BCG translating complex technology concepts into narratives for business audiences. His other writing includes From Tools to Systems and Rules vs. Tools.