Bard was an experimental conversational AI service from Google, announced on February 6, 2023 and opened to public access in the United States and the United Kingdom on March 21, 2023. Google described it at both points as "an early experiment that lets you collaborate with generative AI," powered by "a lightweight and optimized version of LaMDA," Google's Language Model for Dialogue Applications (Bard launch announcements (Google, February–March 2023)).
| Field | Value |
|---|---|
| Developer | Google (Google DeepMind) |
| Announced | February 6, 2023 (trusted testers) |
| General access | March 21, 2023 (US and UK) |
| Underlying model | Lightweight, optimized version of LaMDA |
| Parameters | Not disclosed |
| Open weights | No |
| Published benchmarks | None in the launch announcements |
Release sequence and rationale
The February 6 announcement, under Sundar Pichai's byline, opened Bard "to trusted testers ahead of making it more widely available to the public in the coming weeks." Google stated the initial release used its lightweight LaMDA version because the "much smaller model requires significantly less computing power, enabling us to scale to more users, allowing for more feedback" (Bard launch announcements (Google, February–March 2023)). The March 21 post by Sissie Hsiao and Eli Collins opened access in the US and UK and gave the same evaluative rationale: "the next critical step in improving it is to get feedback from more people."
Both posts frame Bard as combining "the breadth of the world's knowledge" with Google's large language models, drawing "on information from the web to provide fresh, high-quality responses."
Stated limitations
Google's March 21 post is unusually explicit about failure modes for a product announcement. It states that because LLMs "learn from a wide range of information that reflects real-world biases and stereotypes, those sometimes show up in their outputs," and that "they can provide inaccurate, misleading or false information while presenting it confidently." The post illustrates the point with an error in its own screenshot — Bard gave the wrong scientific name for the ZZ plant, which Google corrected in the caption to Zamioculcas zamiifolia (Bard launch announcements (Google, February–March 2023)). See Sycophancy and Hallucination.
The mitigation Google described was presentational rather than technical: offering the user a choice among several drafts of each response, with the option to ask follow-ups or request a retry.
Accompanying commitments
The February announcement bundled three further items with the Bard release. Google stated that "AI-powered features in Search that distill complex information and multiple perspectives into easy-to-digest formats" would begin rolling out "soon," positioning the change around queries where "there's no one right answer." It announced that developer onboarding for a Generative Language API, "initially powered by LaMDA with a range of models to follow," would begin the following month. And it named Google Cloud partnerships with Cohere, C3.ai and Anthropic, the last "just announced last week" (Bard launch announcements (Google, February–March 2023)).
Open questions
- Parameter counts, context length, training data and evaluation results were not disclosed for either the trusted-tester or general-access release.
- Bard's subsequent renaming and its succession by later Google assistant products are not documented here; no source covering that transition has been ingested.
Relationships
- related: Google DeepMind — the developer.
- related: Bard launch announcements (Google, February–March 2023) — the two launch announcements.
- related: Claude 1 (Anthropic) — a competing assistant launched five weeks later.
- related: Sycophancy and Hallucination — the failure mode Google concedes and demonstrates in the launch post.