The new AI era: A guide for non-tech businesses

Hello folks,

If you're not yet thinking about how artificial intelligence (AI) fits into your company's strategy, it's time to start. We are in the midst of what I call an AI "supercycle", following previous tech revolutions such as the internet, cloud computing, and mobile tech. Now, it's AI's turn to reshape our world, from boardrooms to break rooms, and beyond.

You might think AI is still a distant, abstract concept. Not so fast. Machine learning (ML) is already subtly shaping your day-to-day life. Whether it's Google's ad targeting, Uber's pricing, or Snapchat's filters, AI has penetrated every digital nook and cranny. Even behind the scenes, it's been hard at work detecting spam emails, preventing credit card fraud, and moderating content.

And now? AI research is accelerating. Pioneering models like GPT-4, DALL-E, and LaMBDA are unlocking unprecedented capabilities, drawing upon colossal amounts of training data and the 'magic' of prediction.

AI's evolution has been spurred by what's known as "transformers", introduced in a landmark 2017 paper by Google Brain. Put simply, transformers help AI focus on what's essential in its input. This enhanced attention mechanism has paved the way for AI to be more fluent, more scalable, and capable of learning in-context.

This progress has been disrupting industries and grabbing consumer attention left, right, and center. We're seeing AI startups appearing like mushrooms after rain, each offering fresh takes on how AI can enhance productivity, consumer experiences, and more.

Now we're faced with three big questions:

What changes in this new AI era?
Who profits from these changes?
How do content publishers fit into the picture?

To the first question, it's clear that many industries are on the brink of transformation. AI is changing the way we search, learn, write, and code. The legal industry is set to benefit from automated research, while graphic design may shift towards design democratization and personalized marketing.

When it comes to the second question, both incumbents and startups stand to profit. Established businesses can leverage their distribution channels and existing data, while startups can reimagine workflows and embrace risk-taking in a gold rush that draws talent from far and wide.

Regarding the third question, it's a pivotal time for content publishers. The potential lies in more effective re-engagement of audiences and an opportunity to redefine the relationship between publishers, brands, and user-generated content.

So, where do you start with your company's AI strategy? The first step is understanding that 99.9% of AI products will be built by engineers, not researchers or data scientists. We're in an age of increasingly rich model/tool ecosystems, which reduces the need to build from scratch.

Your company's AI strategy should progress through four levels: adding AI to your current offerings, automating workflows, replacing existing workflows with AI-powered ones, and eventually, doing the work for your customers. And remember, if you don't take these steps, eager startups are waiting in the wings.

Considerations for implementing AI include the cost of the model (including potential vendor lock-in, training, data annotation, and feedback), alignment (balancing creativity and error-handling), and managing challenges like hallucinations, safety, and bias.

While it's too early to draw any firm conclusions about the full impact of the AI supercycle, one thing is clear: AI is the single largest value creation opportunity of our lifetimes. There are no known near-term limits to large language model performance improvement. We're in a race to incorporate AI into products and upgrade user experiences. It's an incredibly exciting time in technology. So, let's make sure we seize this opportunity and be part of shaping a more intelligent future.

‍