Artificial Intelligence
Study finds partner content gets priority in ChatGPT
An analysis said ChatGPT handled partner pages as prepared summaries and ranked them ahead of ordinary web results, while later research found uneven treatment and imperfect citation.
- The article said ChatGPT did not fetch partner pages live but received them as prepared summaries.
- It said ordinary web pages arrived as raw text snippets, and partner entries were placed ahead of other web results.
- According to Konitzny's analysis of 1,000 cases, summaries named 70 per cent of entries in positions 1 to 3 and 23 per cent in positions 7 to 10.
- The article said OpenAI had promised partners more prominent placement in 2024.
- It also said a later study from Hamburg found Springer media appearing more often in the web interface than through the API.
The article said ChatGPT did not fetch partner pages live. Instead, it received prepared summaries, while ordinary pages came in as plain snippets, and partner entries were placed ahead of other results.
According to Konitzny's analysis of 1,000 cases, summaries named 70 per cent of entries in positions 1 to 3. In positions 7 to 10, they named 23 per cent.
The piece said OpenAI had promised partners more prominent placement in 2024. It also said a later Hamburg study found Springer media appearing more often in the web interface than through the API, and noted that citations were not always correct.
Named in this story
People
- Konitzny
- analysed 1,000 cases
Companies
- OpenAI
- was said to have promised partners more prominent placement
Organisations
- Springer media
- was found to appear more often in the web interface than through the API
Products and systems
- ChatGPT
- received partner content and web results
Places
- Hamburg
- was where the later study was based
How the source tells it
The register was analytical but used a sharp, comparative framing to imply unequal treatment.
- unfairness a loaded contrast between privileged partner summaries and ordinary web snippets