Appen Research

@AppenResearch

Human data for frontier AI. Research and insights from Appen.

Joined May 2026

210

Appen Research
@AppenResearch
6h
Coding agents need real enterprise environments rather than synthetic tasks. Appen provides 30+ private coding environments with 25M+ lines of licensed, previously unindexed code, ~58.8% test coverage and human-written unit tests built for realistic agent training and
Appen Research
@AppenResearch
7h
Multimodal hallucination is structurally harder to catch than text hallucination. A text hallucination can be checked against an external source. A visual one can only be verified against the source image, which is why it survives at scale. Featured by @TheAIInnov,
Appen Research
@AppenResearch
Aug 13
Healthcare AI may become data-bound before it becomes model-bound. Clinical models increasingly need multimodal, real-world data: imaging + reports, biosignals, multi-turn clinical dialogue, tool-use trajectories and expert-validated reasoning. That data doesn’t scale like web
Appen Research
@AppenResearch
Aug 11
Can LLMs actually reason, or are we increasingly talking about systems that retrieve, plan, call tools and invoke specialized solvers? In the latest episode of The Data Layer podcast, @Oracle Chief AI Scientist Dan Roth @DanRothNLP joins Appen to dig into reasoning, retrieval,
Appen Research
@AppenResearch
Aug 3
"Bring 100 experts into the room, and you'll get 100 different opinions." At SlatorCon London 2026, Appen's @SergioBrucc discussed why domains like coding, STEM, healthcare, legal, and HR require expert judgment, and why expertise must span cultures, languages, and modalities.