We have Google Chrome Privacy Sandbox, The Trade Desk’s Unified ID 2.0 and joint projects with companies like LiveRamp and Epsilon that created identifiers like RampID, and Core ID, respectively.

These workarounds for the loss of third-party targeting cookies in browsers are intended to ease the transition to the next phase of ad targeting. It has been painful for many advertisers, but generative artificial intelligence (GAI) technology will eventually step in to support the next phase of ad targeting.

Generative AI will allow companies to build highly personalized consumer profiles with just a few bits of consumer data to target ads and follow behavioral patterns. It will become known as the collection of GAI data for advertising. 

For some, targeting technologies (I do not have a crystal ball — just making an educated observation) will in part come from the development of small language models embedded in devices like mobile phones, tablets and laptops.

Microsoft Research this week announced the release of Phi-2, the company’s language model, which is small enough to run on a laptop or mobile device.

The company says it demonstrates “outstanding reasoning and language understanding capabilities.” Its compact size will enable Microsoft to embed it into small devices.

For now, Microsoft has made Phi-2 available in the Azure AI Studio model catalog to help developers with research and development on language models.

Microsoft says Phi-2 — with 2.7 billion parameters of connections between artificial neurons — improves performance, compared to much larger models such Meta’s Llama 2-7B with 7 billion parameters, and Paris-based startup Mistral-7B, another 7 billion parameter model.

The key, which I cannot emphasize enough, is that technology will fit into small devices.

Gemini — a series of Google’s language models — was announced earlier this month. Beginning today, developers and enterprise customers have access to Gemini Pro via the Gemini API in Google AI Studio or Google Cloud Vertex AI.

Google also has been working on Project Elmann, a technology based on GAI that would see all the data and photographs in a user’s Google account, from notes to photos and location data. The technology will have the ability to “answer previously impossible questions.”

Elmann will present a person’s life in chapters and can display content from school years or a specific vacation — data that would help Google serve and target ads.

The intelligence will have the ability to pick up behavioral patterns. An Ellmann chatbot will allow users to ask and receive answers from questions about their life. 

Google named the project after the American literary critic Richard David Ellmann.