Home / Frontiers

AI for PV

For all readersPlanned

AI in PV R&D, manufacturing and O&M

Articles

• Perovskite Literature AI Assistants Are Only "Half Right": The Real Weakness Is Not the Model but the Data TableA new arXiv preprint uses 755 quantitative questions and 45,441 measured device records to score "perovskite literature AI assistants": zero-shot answers reach only 1.6–6.5%, conventional passage retrieval at most 16–23%, and only when the model queries a structured device table with read-only SQL does accuracy rise to about 51%. We briefly review the work and discuss what it means for PV R&D and technology-transfer engineers using AI for literature search, and for companies building internal experimental databases.
• Build a "Self-Learning" EL Defect Detection AI on Your Own ComputerEL image judgement relies on human eyes, standards are hard to align, and experience leaves with people. Using open-source YOLO on an ordinary computer, we build an EL defect detection program for cells and modules from scratch, with a "detect, human review, relearn, exam-based selection" loop that makes the model more accurate with use. Key code and a pitfall checklist included.
• Read the IEC 61215 Test Sequence in Ten Minutes with AIFour steps, four prompt templates and one sequence map for reading the IEC 61215:2021 test flow chart with AI in ten minutes: what each of the seven branches simulates, what the two gates judge, the logic behind the order, the main changes in the 2021 edition, and the three places AI most often gets wrong.
• Five of the Most Useful AI Prompts for PV EngineersFive ready-to-use AI prompt templates for failure analysis, standards interpretation, experiment design, data analysis and technical reports. We explain the five things a good prompt covers, why each template is written the way it is, how to follow up, and three red lines for using AI.
• AI Is Here: Will PV Technical Jobs Be Replaced?AI replaces tasks, not jobs. This article takes six PV technical roles (materials, module design, process, reliability, failure analysis and certification) one by one to see which work is losing value and which is gaining it, plus three hurdles AI cannot get around and five pieces of advice for practitioners.