Lactic Acid Related Molecule Reaction Outcome Prediction via AI Chatbots
- 1Carmel High School, 520 E Main St, Carmel, IN, United States, 46032
Abstract
Lactic acid (LA) and its related molecules like its polymer (Polylactic acid) have large importance in organic chemistry reactions. The objective of this experimental study was to determine the effectiveness of using AI chatbots to predict the outcomes and key findings of chemical reactions associated with the lactic acid chemical, along with understanding which chatbots are more effective to use in order to investigate a more efficient method to form hypotheses about these reactions. Reaction conditions of chosen reactions were put into specific models of ChatGPT, Perplexity, Copilot, and Gemini to test whether the outcomes they predicted matched with the key findings in the research papers the reactions were sourced from. Using statistical analysis by Cochran’s Q Test (p = 0.247237), a statistically significant difference between types of chatbot used for product prediction was not found. However, a Wilson Score Interval of the Perplexity Pro default model with 90% confidence yielded an interval of (0.6010, 0.8847). The findings essentially suggest a relatively high accuracy interval for the proportion of the types of LA related reactions in this study that the Perplexity Pro default model can predict correctly. The implications of this emphasize the importance of further examining what reactions can and cannot be used in AI chatbots to predict the outcomes.
Related
- IMPLICIT LEARNING: REACTION OUTCOME PREDICTION FROM POSITIVE-ONLY DATA
- Single-molecule cold chemistry: Isotopic branching ratios of the H2O+ + HD reaction at cryogenic temperatures
- Isomer-resolved ion-molecule reaction kinetics of HCO+ and HOC+
- Times are changing but order matters: transferable prediction of small-molecule liquid chromatography retention times
- Bridging the Literature-to-Laboratory Gap with Language Models and Active Learning for Reaction Feasibility Prediction and Autonomous Exploration
BibTeX
@misc{sahu2026lactic,
title = {{Lactic Acid Related Molecule Reaction Outcome Prediction via AI Chatbots}},
author = {Atharva Sahu},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15009700/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009700/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Lactic Acid Related Molecule Reaction Outcome Prediction via AI Chatbots
AU - Atharva Sahu
PY - 2026
DA - 2026/09/30
PB - ChemRxiv
DO - 10.26434/chemrxiv.15009700/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15009700/v1
AB - Lactic acid (LA) and its related molecules like its polymer (Polylactic acid) have large importance in organic chemistry reactions. The objective of this experimental study was to determine the effectiveness of using AI chatbots to predict the outcomes and key findings of chemical reactions associated with the lactic acid chemical, along with understanding which chatbots are more effective to use in order to investigate a more efficient method to form hypotheses about these reactions. Reaction conditions of chosen reactions were put into specific models of ChatGPT, Perplexity, Copilot, and Gemini to test whether the outcomes they predicted matched with the key findings in the research papers the reactions were sourced from. Using statistical analysis by Cochran’s Q Test (p = 0.247237), a statistically significant difference between types of chatbot used for product prediction was not found. However, a Wilson Score Interval of the Perplexity Pro default model with 90% confidence yielded an interval of (0.6010, 0.8847). The findings essentially suggest a relatively high accuracy interval for the proportion of the types of LA related reactions in this study that the Perplexity Pro default model can predict correctly. The implications of this emphasize the importance of further examining what reactions can and cannot be used in AI chatbots to predict the outcomes.
ER -