Google reveals Bard to take on ChatGPT

Watching ChatGPT capture the world's imagination, Google finally pressed the accelerator.

Google announced Bard, an experimental conversational AI service powered by its LaMDA model. It is rolling out to trusted testers before opening widely to the public in the coming weeks.

At the same time, Google confirmed that generative AI features will soon appear directly inside Google Search to distill complex questions.

Lightweight LaMDA and a revamped Search

To manage computing power and scale fast, Google is launching Bard with a lightweight version of LaMDA.

This smaller model requires significantly less compute, allowing Google to collect more feedback from a larger user base. Bard draws from web data to answer queries, such as breaking down discoveries from NASA’s James Webb Space Telescope for a nine-year-old.

Beyond chat, Google Search itself is changing. Instead of simple factual lookups like the number of piano keys, users increasingly ask questions with no single right answer, such as whether learning piano or guitar is easier.

Google will soon deploy AI features at the top of Search that synthesize diverse perspectives into easy-to-read formats.

For developers, Google plans to introduce a Generative Language API next month, initially powered by LaMDA. Google Cloud is also backing this ecosystem through compute partnerships with Cohere, C3.ai, and Anthropic.

Paris reveals and a painful stock drop

Two days later at an event in Paris, Google expanded on its vision for Search and Maps.

It highlighted multisearch, letting users search with photos and text together, and teased an upcoming "multisearch near me" tool. Google Maps also received Immersive View, a feature powered by neural radiance fields (NeRF) that merges aerial and Street View imagery into interactive 3D landmarks.

However, the sudden rush to ship backfired quickly.

During a product demo, Bard generated an inaccurate fact regarding the James Webb Space Telescope. Despite debate over whether the answer held truth in certain contexts, Alphabet shares sank more than 8% in a single day.

To make matters more intense, Microsoft unveiled a ChatGPT-powered preview of its Bing search engine on the exact same day Google announced Bard.

Why it matters

We think this is the moment Google’s two-decade search monopoly officially entered a high-stakes turf war.

For everyday users, Bard itself is not yet widely accessible, but the upcoming AI summaries in Search could redefine how we explore open-ended topics. The trade-off is precision. As the demo blunder showed, hallucinations in search results carry immediate real-world consequences.

Google’s move to run a lightweight LaMDA model is a smart, pragmatic infrastructure choice. Yet it highlights the immense cost pressure of running generative models at search engine volume.

Ultimately, Google finds itself in a classic innovator’s dilemma. Summarizing answers directly threatens the blue links that drive its core search advertising engine. Microsoft can afford to be reckless with Bing, but Google has to reinvent its most lucrative machine while defending it from rapid disruption.