Generative AI (GenAI)
Generative AI is the umbrella term for AI systems that create new content — text, image, audio, video or code — based on trained models rather than retrieving existing information.
Generative AI differs from classical predictive AI: where traditional models classify or predict, generative models create something new. ChatGPT, Claude, Midjourney, Sora and GitHub Copilot are all generative AIs built on foundation models with billions of parameters. For marketers this means a fundamentally different interaction with search (AI Overviews, Perplexity) and content creation (drafts, variants, personalisation).
Example
An e-commerce site uses GenAI to personalise 5,000 product descriptions per audience. Input: product + audience segment. Output: unique, brand-consistent description. Time saved: 400 writing hours.
Frequently asked questions
How does GenAI differ from traditional AI?
Traditional AI predicts, classifies or optimises within existing data (fraud detection, recommendations). GenAI creates new output: text, image, audio, code.
Which models count as GenAI?
LLMs (GPT, Claude, Gemini), image models (DALL-E, Midjourney, Stable Diffusion), video (Sora, Runway), audio (ElevenLabs), and multimodal models combining these.
Related terms
Further reading
- → Our service: GEO