# Generation Probabilities Are Not Enough: Uncertainty Highlighting in AI Code Completions arXiv: https://arxiv.org/abs/2302.07248 ยท HTML: https://mlsanity.com/paper/2302.07248/ ## Key facts - Cited by 20 papers in the map (data through 14 Oct 2026). - Builds on 25 of its 75 references that are in the map. - First posted on arXiv 14 Feb 2023. ## Builds on (most cited first, 25 of 25) - [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017) - [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020) - [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017) - [Survey of Hallucination in Natural Language Generation](https://mlsanity.com/paper/2202.03629/) (2022) - [Language Models (Mostly) Know What They Know](https://mlsanity.com/paper/2207.05221/) (2022) - [Teaching Models to Express Their Uncertainty in Words](https://mlsanity.com/paper/2205.14334/) (2022) - [How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answering](https://mlsanity.com/paper/2012.00955/) (2020) - [Asleep at the Keyboard? Assessing the Security of GitHub Copilot's Code Contributions](https://mlsanity.com/paper/2108.09293/) (2021) - [Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance](https://mlsanity.com/paper/2006.14779/) (2020) - [Self-critiquing models for assisting human evaluators](https://mlsanity.com/paper/2206.05802/) (2022) - [Effect of Confidence and Explanation on Accuracy and Trust Calibration in AI-Assisted Decision Making](https://mlsanity.com/paper/2001.02114/) (2020) - [Reducing conversational agents' overconfidence through linguistic calibration](https://mlsanity.com/paper/2012.14983/) (2020) - [Do Users Write More Insecure Code with AI Assistants?](https://mlsanity.com/paper/2211.03622/) (2022) - [Uncertainty as a Form of Transparency: Measuring, Communicating, and Using Uncertainty](https://mlsanity.com/paper/2011.07586/) (2020) - [Productivity Assessment of Neural Code Completion](https://mlsanity.com/paper/2205.06537/) (2022) - [Reading Between the Lines: Modeling User Behavior and Costs in AI-Assisted Programming](https://mlsanity.com/paper/2210.14306/) (2022) - ["Why is 'Chicago' deceptive?" Towards Building Model-Driven Tutorials for Humans](https://mlsanity.com/paper/2001.05871/) (2020) - [Help me write a poem: Instruction Tuning as a Vehicle for Collaborative Poetry Writing](https://mlsanity.com/paper/2210.13669/) (2022) - [Investigating Explainability of Generative AI for Code through Scenario-based Design](https://mlsanity.com/paper/2202.04903/) (2022) - [What is it like to program with artificial intelligence?](https://mlsanity.com/paper/2208.06213/) (2022) - [Opal: Multimodal Image Generation for News Illustration](https://mlsanity.com/paper/2204.09007/) (2022) - [Teaching Humans When To Defer to a Classifier via Exemplars](https://mlsanity.com/paper/2111.11297/) (2021) - [Algorithmic Risk Assessments Can Alter Human Decision-Making Processes in High-Stakes Government Contexts](https://mlsanity.com/paper/2012.05370/) (2020) - [Show or Suppress? Managing Input Uncertainty in Machine Learning Model Explanations](https://mlsanity.com/paper/2101.09498/) (2021) - [How Readable is Model-generated Code? Examining Readability and Visual Inspection of GitHub Copilot](https://mlsanity.com/paper/2208.14613/) (2022) ## Built on by (most cited first, 20 of 20) - [Sociotechnical Safety Evaluation of Generative AI Systems](https://mlsanity.com/paper/2310.11986/) (2023) - [From Calibration to Collaboration: LLM Uncertainty Quantification Should Be More Human-Centered](https://mlsanity.com/paper/2506.07461/) (2025) - [QE4PE: Word-level Quality Estimation for Human Post-Editing](https://mlsanity.com/paper/2503.03044/) (2025) - [Surfacing Variations to Calibrate Perceived Reliability of MLLM-generated Image Descriptions](https://mlsanity.com/paper/2507.15692/) (2025) - [Human-AI Experience in Integrated Development Environments: A Systematic Literature Review](https://mlsanity.com/paper/2503.06195/) (2025) - [Unsupervised Word-level Quality Estimation for Machine Translation Through the Lens of Annotators (Dis)agreement](https://mlsanity.com/paper/2505.23183/) (2025) - [The Impact of LLM-Assistants on Software Developer Productivity: A Systematic Review and Mapping Study](https://mlsanity.com/paper/2507.03156/) (2025) - [Statistical Methods in Generative AI](https://mlsanity.com/paper/2509.07054/) (2025) - [Cognitive Spillover in Human-AI Teams](https://mlsanity.com/paper/2407.17489/) (2024) - [From Consumption to Collaboration: Measuring Interaction Patterns to Augment Human Cognition in Open-Ended Tasks](https://mlsanity.com/paper/2504.02780/) (2025) - [Localized Calibrated Uncertainty in Code Language Models](https://mlsanity.com/paper/2512.24560/) (2025) - [Overseeing Agents Without Constant Oversight: Challenges and Opportunities](https://mlsanity.com/paper/2602.16844/) (2026) - [Functional Entropy: Predicting Functional Correctness in LLM-Generated Code with Uncertainty Quantification](https://mlsanity.com/paper/2605.28500/) (2026) - [HiLDe: Intentional Code Generation via Human-in-the-Loop Decoding](https://mlsanity.com/paper/2505.22906/) (2025) - [Effects of Generative AI Errors on User Reliance Across Task Difficulty](https://mlsanity.com/paper/2604.04319/) (2026) - [Beyond Autocomplete: Designing CopilotLens Towards Transparent and Explainable AI Coding Agents](https://mlsanity.com/paper/2506.20062/) (2025) - [Does In-IDE Calibration of Large Language Models work at Scale?](https://mlsanity.com/paper/2510.22614/) (2025) - [Fine-grained Approaches for Confidence Calibration of LLMs in Automated Code Revision](https://mlsanity.com/paper/2604.06723/) (2026) - [Choose Your Own Adventure: Non-Linear AI-Assisted Programming with EvoGraph](https://mlsanity.com/paper/2604.18883/) (2026) - [Human agency in initial human-AI proof formalization workflows](https://mlsanity.com/paper/2606.04273/) (2026) ## Abstract (arXiv) Large-scale generative models enabled the development of AI-powered code completion tools to assist programmers in writing code. However, much like other AI-powered tools, AI-powered code completions are not always accurate, potentially introducing bugs or even security vulnerabilities into code if not properly detected and corrected by a human programmer. One technique that has been proposed and implemented to help programmers identify potential errors is to highlight uncertain tokens. However, there have been no empirical studies exploring the effectiveness of this technique -- nor investigating the different and not-yet-agreed-upon notions of uncertainty in the context of generative models. We explore the question of whether conveying information about uncertainty enables programmers to more quickly and accurately produce code when collaborating with an AI-powered code completion tool, and if so, what measure of uncertainty best fits programmers' needs. Through a mixed-methods study with 30 programmers, we compare three conditions: providing the AI system's code completion alone, highlighting tokens with the lowest likelihood of being generated by the underlying generative model, and highlighting tokens with the highest predicted likelihood of being edited by a programmer. We find that highlighting tokens with the highest predicted likelihood of being edited leads to faster task completion and more targeted edits, and is subjectively preferred by study participants. In contrast, highlighting tokens according to their probability of being generated does not provide any benefit over the baseline with no highlighting. We further explore the design space of how to convey uncertainty in AI-powered code completion tools, and find that programmers prefer highlights that are granular, informative, interpretable, and not overwhelming.