Skip to main content

Generative AI in non-technical terms

What do playing chess and forecasting the weather have in common? To do either of these things, humans must learn to recognize and predict patterns in a way that's analogous to machine learning.

The more a person plays chess, the more they start to recognize common opening sequences, patterns in opponents' strategies, and positional patterns that lead to good or bad outcomes. Their brain essentially builds models that allow them to better predict the best moves and what their opponent might do in response.

Meteorologists forecast weather patterns based on recognizing signatures in data from satellites, radars, etc. that match situations they've encountered before. Their brains have registered relationships between patterns in the data to patterns in how weather systems tend to evolve.

Artificial intelligence has developed in a very similar way.

Machine learning: The start of something good

Long before generative AI was making headlines, computer scientists developed machine learning (ML). The idea of machine learning first showed up in the late 1950s, yet practical applications didn’t really surface until the ‘80s and ‘90s.

Machine learning is an approach that allows computers to learn by recognizing patterns and signals without being explicitly programmed or instructed by a developer. Think of a bank wanting to identify credit card fraud. Before machine learning, the bank would ask a developer to create an application that uses rules to detect fraudulent transactions. But credit card fraud can occur in so many different ways, it would be near impossible to come up with all the different ways and write rules for each occurrence.

Now, with machine learning, banks can use their own historical dataset containing countless credit card transactions, each labeled “fraudulent” or “legitimate.” This data is fed into an algorithm which starts to identify patterns and cues in the data, like purchase amount or timing and location of purchases.


Algorithms can learn which transactions are more likely to be fraudulent.


Just like how you know to reach for an umbrella when you see a cloudy sky, the bank’s algorithm uses these signals and patterns to identify which transactions are more likely to be fraudulent. (Cloudy sky = possible rain, unusually large purchase amount = possible fraud.)

Teaching computers to learn from past data and make predictions on new data was a big step leading us towards generative AI.

Deep learning: Understanding nuances

As machine learning became more commonplace, researchers looked for ways to make it even smarter. Inspired by the human brain, they created artificial "neural networks" which could recognize very complex patterns in large amounts of data, like images, speech, and video.

How does this differ from the traditional machine learning pattern recognition? Deep learning models can detect patterns and signals at an even more granular level.


Deep learning can always spot a dog, even if it’s in disguise.


Rather than just memorizing rules and cues given in training, deep learning can learn to understand the concept of “dogness.” And that makes deep learning more accurate, flexible, and intelligent when it comes to making judgements, just like a human brain.

NLP: Learning to speak like a human

Deep learning and neural networks also revolutionized natural language processing (NLP). NLP is when computers understand, interpret, and generate human languages and it’s used for things like translation, speech recognition, and chatbots.

Deep learning NLP analyzes unimaginable amounts of language data from sources like Wikipedia, books, news articles, and scientific journals. It then learns the intricate patterns around how sentences and words relate, all without needing explicit programming. This creates more human-like output of translations, summarizations, and conversations that early NLP could not achieve.

Search engines are a great example of this. They use NLP, in addition to past search history and searcher intent, to serve up what they deem to be the best results. They may even suggest search terms as you type, based on popular searches.

Generative AI: The next generation

Machine learning, especially approaches like neural networks and deep learning, advanced to the point where computers could recognize and label content in images, text, and audio with great accuracy. For the next stage, researchers wanted to move past labeling content and start creating original content.

This is how we got generative AI models. These models are trained on data, just like deep learning models, but produce new synthesized data as output. They create completely original images, text, audio, video, or code that are similar to the training data.


The more data generative AI is exposed to, the more it has to draw upon and learn from to create new content.


The process is a little like the way young children make their first artworks. They begin to recognize shapes and colors based on the books adults read to them. Eventually, they’ve seen enough that they know they can draw a circle, color it yellow and it will resemble the sun. The more they practice—and the more memories they have to draw upon—the more creative their artwork becomes.

Practical uses: Generative AI in business

All of these advancements are unlocking new opportunities to reduce costs, increase revenue, and mitigate risk across the organization. Generative AI specifically has the potential to automate repetitive tasks, uncover actionable insights, and enhance productivity. But what does that mean in practical terms?

  • Generative AI-powered chatbots and virtual assistants, organizations are boosting customer satisfaction and loyalty while reducing support costs. These tools can provide 24/7 coverage and hyper-personalized recommendations tailored to each customer.
  • Generative AI can pore through vast datasets to detect patterns and generate summaries. This reduces the analytical burden on employees and allows them to spend time on higher-value strategic initiatives. Key insights uncovered can inspire innovative new products and marketing campaigns grounded in data.
  • For developers and IT teams, generative AI speeds up coding by suggesting ready-made blocks of code. This structured code is clearly commented for future maintenance and collaboration. By cutting development timelines and costs, resources are freed up to focus on novel applications that drive the business forward.

There are just a few examples of how generative AI can automate work so employees can maximize their potential and employers can meet their objectives.

Taking the next step

Generative AI was not an overnight success story. It’s the culmination of years of foundational research and innovation in machine learning and neural networks. Now, these powerful models are being used and providing tremendous value in organizations around the world.

However, fully realizing the potential of generative AI requires strategic investments in talent, data, and computing infrastructure. Executives must have the vision and conviction to provide the resources needed to develop and deploy this technology responsibly.

The opportunity is clear - generative AI can:

  • drive improvements in efficiency, insight, and competitive differentiation.
  • automate repetitive workflows, reallocating employees to more meaningful work.
  • reveal trends and opportunities hidden within massive datasets.
  • can create on-demand, personalized content and experiences for each customer.

Now is the time to plan for the many ways that generative AI will transform your organization—from building a culture of data, to upskilling and hiring staff, and making space for innovation—how will you make generative AI your competitive advantage?