Can Google’s Magi AI Search Beat Bing and Keep Samsung as a Partner?

Google’s core revenue comes from advertising tied to its search engine, and the rise of AI‑enhanced search from competitors like Microsoft’s Bing, which now leverages OpenAI’s GPT‑4, has put pressure on the company.

In response, Google launched the internal “Magi” project, mobilizing over 160 designers, engineers, and executives to create a more personalized, AI‑driven search experience for users on phones and computers.

Early testing of Magi was accelerated after Samsung hinted at adopting Bing as the default search engine on its Galaxy devices, threatening Google’s $30 billion annual payment to Samsung for search‑service licensing.

Google plans to roll out Magi initially in the United States, targeting one million users at launch and expanding to 30 million by year‑end, while also overhauling its existing search architecture.

The new AI search aims to understand user interests, provide contextual answers, generate code, create images, and even assist with language learning, using tools such as GIFI (image generation), Tivoli Tutor (AI‑driven language teaching), and Searchalong (real‑time AI assistance while browsing).

Challenges include ensuring content quality, protecting user privacy, and maintaining fairness and accountability of AI models.

Google’s leadership emphasizes a cautious rollout, noting that not every brainstormed idea will become a product, and stresses the need for strong regulation to mitigate risks like deep‑fake videos.

Overall, Magi represents Google’s effort to keep its search dominance by integrating generative AI features that are conversational, personalized, and capable of handling complex tasks such as transactions and code generation.

AI‑enhanced image results via GIFI

Open‑ended language tutoring with Tivoli Tutor

Real‑time browsing assistance through Searchalong