Thinking of the future - with Meta's Thought Preference Optimisation

Imagine your marketing team brainstorming hundreds of genuinely inspired ideas in seconds.

That's the potential of Thought Preference Optimisation (TPO), a new AI technology from Meta.

TPO is like giving your AI a brain. Unlike standard AI, which rushes to spit out the first answer, TPO takes time to 'think' things through. It generates internal thoughts, invisible to you, and uses them to refine its responses before presenting the best one.

This is a big development for marketing. In trials, TPO outperformed standard AI on key marketing and creative writing benchmarks.

Think of it as your AI team going through rounds of internal brainstorming and presenting you with only the most polished, effective ideas.

Here's a quick example:

Prompt: Write a tagline for a new brand of sustainable sneakers made from recycled ocean plastic.

Standard LLM Output:

Walk with the ocean.

TPO Output:

1. Internal Thoughts:

The tagline should be concise and impactful.

It should highlight the sustainability aspect.

It should appeal to the target audience who care about the environment and style.

Possible themes: ocean conservation, recycled materials, stylish and eco-conscious.

2. Tagline: Style that tides the world.

The standard LLM output is quick and direct, but lacks depth and creativity. It fails to convey the unique selling proposition.

The TPO output demonstrates a more deliberate and thoughtful process. The internal thoughts reveal how the AI considers different factors and explores potential themes before arriving at a tagline that is more creative, memorable, and relevant to the brand's mission.

When can you get your hands on this? While still under development, TPO builds on the latest AI research from Meta and OpenAI.

Find out more on the latest episode of AI Today...