Research AI Landscape

google-gemini

3 min read

Google Gemini

Status: trial | Maturity: production-ready | Last reviewed: 2026-03-20

What it does

Google Gemini is Google’s family of large language models, available as a consumer/institutional assistant at gemini.google.com and integrated into Google Workspace (Docs, Sheets, Gmail, Drive) as “Gemini for Workspace”. The models (Gemini 2.0 Flash, 1.5 Pro, 1.5 Ultra/Advanced) support text, image, audio, and video input (multimodal).

For researchers, Gemini offers capabilities for drafting, summarising, analysing data, coding assistance, and general intellectual tasks. The Deep Research feature (Gemini Advanced) performs multi-step web research and produces structured reports.

Key capabilities

  • General-purpose AI assistant: writing, summarising, explaining, analysing
  • Multimodal inputs: text, images, documents, audio
  • Gemini for Workspace: integrated into Google Docs, Sheets, Slides, Gmail
  • Deep Research: autonomous multi-step web research and report generation
  • NotebookLM integration (separate tool, already catalogued)
  • Long context window (1 million tokens in 1.5 Pro) for large document analysis
  • Code generation and explanation

Fit for Sheffield researchers

Sheffield provides Google Workspace to staff and students — Gemini is the AI layer in that ecosystem. The existing-university-license flag is set to true because Sheffield’s Google Workspace provides base Gemini access; confirm exact feature availability with CiCS (some Advanced features may require additional licence tier).

Particularly relevant for: grant writing assistance, literature synthesis, drafting, data analysis in Sheets, and multimodal document analysis. NotebookLM (already catalogued as adopt) is the recommended tool for document-grounded research; Gemini is the better choice for open-ended generation tasks.

Licensing and access

  • Gemini (base): free at gemini.google.com; available within Google Workspace institutional licence
  • Gemini for Workspace: depends on Workspace licence tier (Business Standard/Plus, Enterprise)
  • Gemini Advanced: Google One AI Premium (~£19/month individual) or Workspace add-on
  • Confirm Sheffield’s current Workspace tier with CiCS/Library

Trust and governance notes

Google Workspace Enterprise-tier data is not used to train AI models (per Google’s Workspace data processing terms). Researchers using personal/free Gemini accounts should review Google’s privacy policy regarding training data opt-out. Data residency is US. GDPR compliance status is unclear pending confirmation of Sheffield’s Workspace tier and Google’s applicable terms.

Do not use Gemini (or any general-purpose LLM) to process personal data, confidential research data, or commercially sensitive information unless the appropriate data processing agreement is in place. Confirm with the University’s Information Governance team for sensitive use cases.

Sheffield’s AI use guidelines should be followed; disclose AI-assisted content in publications and grant applications per funder requirements.

Observed limitations

  • General-purpose model — less specialised than domain-specific tools for expert tasks
  • Hallucination risk for factual claims, especially in fast-moving research areas
  • Deep Research feature quality is variable — verify sources independently
  • US data residency is a concern for sensitive data
  • AI use disclosure requirements vary by journal, funder, and institution — researchers must check

notebooklm · chatgpt-edu · google-co-scientist · elicit

Sources and evidence

Graph View

Backlinks