How to fix AI Coding Failed on ATLAS.ti
ATLAS.ti 2024 has integrated AI coding, intentional AI coding, AI summaries, and more AI-powered features. This makes thematic analysis of interviews, focus groups, and open-ended question data fast. However, these features may still not be accurate and effective when it comes to analyzing large amounts of data. One common error that pops up on ATLAS.ti is AI Coding Failed: ATLAS ti AI Coding Failed Error occurred during AI coding object reference not set to an instance of an object
One day, I was trying to use AI coding to analyze 10 interview transcripts on ATLAS.ti 2024 software; however, an error popped up saying:
★ ☆ ☆ ☆ ☆
AI Coding Failed
An error occurred during AI Coding:
“This request will cost 500, but you have only left 0”
Your progress has been saved, AI Coding will take it from there on the next run.
I contacted the support, and this is what they said:
★ ★ ★ ★ ★
“Thank you for notifying us. I confirm we have added more AI tokens to your account. You should now be able to continue using ATLAS.ti’s AI tools. We appreciate your understanding. As you know, we strive to always ensure that everyone with an ATLAS.ti license has full access to all our features at no additional cost.
While other software packages provide a limited amount of tokens for AI-based tools and then charge users for additional tokens, we at ATLAS.ti do everything we can to ensure our users always have free access to all of our features and platforms. For most users, the default amount of AI tokens is sufficient, but naturally, there are cases in which a higher volume of data is analyzed, which is why we needed to manually increase the tokens in your account.”
However, I faced the same issue on the software version again. I also tried to free up my seat and log in another time through page: https://my.atlasti.com/licenses, but the issue persisted. What can I do now? I recontacted the ATLAS.ti support, and this is what they said:
★ ★ ★ ★ ★
“The tokens were updated just recently, so it should be working now. Please try applying AI Coding to another document to test it out first, a smaller one preferably”
I uploaded a small-sized document, and finally, the AI coding went back to normal. This problem will be faced whenever one uploads a large-size data. Hence, the error on AI coding can only be solved by chatting with the ATLAS.ti support team who are prompt and professional people.
ATLAS.ti utilizes AI to automatically segment text into manageable chunks, which are then processed by GPT models for repeated analysis. The algorithm combines the analysis results to generate a well-balanced set of codes that cover various topics, ensuring the output is comprehensive yet not excessively large. It is suitable for inductive, deductive and abductive coding.
The process involves analyzing documents segment by segment, typically paragraph by paragraph. Each segment is assessed with a set of questions. When a question is relevant and provides a reasonable answer, this answer is condensed into a code. This code is then assigned to the segment and categorized under the corresponding question’s code category. The codes can be open code, axial codes and In Vivo Codes.
AI Coding does not process very short paragraphs and only analyzes the plain text of documents. Elements such as existing codes, quotations, and formatting do not influence its outcomes. AI coding can be intentional guided by research questions, objectives or hypotheses.
AI coding of interviews, focus groups, and open-ended questions is available on ATLAS.ti for a student license of one month for 10 USD and 6 months for 60 USD.