Imagining what’s possible when machines know everything
Recent events (mostly the release of ChatGPT, followed by Microsoft’s investment in OpenAI, its creator) have highlighted the difference between machine learning and deep learning.
Machine learning is robotic, often failing to grasp context. We might think of it like the brain of an ant. Deep learning is based on artificial neural networks – it mimics the brain. Deep learning is nowhere near as smart as humans yet, but we might imagine it to be like the brain of a guide dog:* awfully good at recognizing patterns and grasping context. Deep learning can be incredibly helpful, just as a guide dog can be incredibly helpful. And since deep learning models run on powerful computers, they are like a million guide dogs.
This analogy has officially gone way too far. But you get the idea.
What is Driving the Conversation?
We talked this week with a few DCA members who described different visions for what AI can do for offers. One laid out a vision for “hyper-personalized offers.” Imagine an ad in which your favorite actor (deep-faked with permission and royalties paid) looks you in the eye, addresses you by name, and suggests that you would love trying a new brand of golf balls, since golf is your passion. He points out that you can get 20% off a sleeve of balls at any pro shop with the Mastercard you already have in your wallet.
It’s worth imagining this. All the technology exists – it just needs to be stitched together and fed with (permissioned) consumer data. When it happens – and at this point, it’s more “when it happens” than “whether it happens” – there will be a mad scramble to adapt to the new reality.
Questions DCA Members are Asking:
- What can we do today to leverage deep learning tools to increase our productivity in touching consumers with compelling offers?
- How much of a first-mover advantage will there be? Are we better off being a fast-follower?
- Does any of this get developed by our company, or do we partner for all of it?
- Do we control any of the input data (i.e., consumer data) that could be critical to creating hyper-personalized offers?
- With whom could we partner to enrich the data we already have?
- How do we deliver increasingly personalized offers without triggering a negative consumer reaction?
Related DCA Resource:
We’re working with a DCA member on the cutting edge of AI to develop an Expert Session on this topic in the coming months – stay tuned!
*A note for cat people. Cats are at least as smart as dogs, and a “guide cat” is laughable only because cat brains are so finely tuned for circumspection and contempt. Please feel free to insert “cat” where you read “dog” above, or to discard the whole idea – with contempt.