Evaluating the Ability of Commercial Search Engines to Help People Answer Health Questions
Bibliographic record
Abstract
The act of seeking information pertaining to medical treatments and self-diagnosis is one of the applications of search engines. However online documents and websites offer convenience and efficiency in accessing information, it is important to acknowledge that they may contain incorrect and also unreliable information, which can potentially lead to adverse consequences such as making harmful medical decisions. This is particularly concerning when search engine users rely solely on the information they encounter through search results, without conducting additional research or seeking guidance from qualified medical professionals. Therefore, it is essential to assess the impact of search engines on users’ behavior and decision-making processes, especially when it comes to health-related decisions. Previous research has been conducted to evaluate the extent to which people may be affected by search engine results when they are responding to health-related questions, upon which our study is based (Pogacar et al., 2017; Ghenai et al., 2020). Their findings indicated that individuals tend to make correct decisions when supplied with a series of correct information as search results, and conversely, they tend to make wrong decisions when presented with a group of search results with incorrect information. The prior research studies used a methodology whereby study participants were presented with static search results, without the ability to actively query a search engine. In our study, we designed and conducted a controlled laboratory study which followed a within-subject design that consisted of presenting a group of participants with 12 topics from TREC 2021 Health Misinformation track with each topic comprising a particular health issue and its corresponding suggested medical treatment. These treatments were categorized as either helpful or unhelpful for each health issue, but the participants were not aware of the true effectiveness of each treatment. The participants were then asked to evaluate the effectiveness of the treatments both with and without utilizing the search engine experience provided to them. The search engine environment was established using modern commercial search engine APIs such as Google and Bing as its underlying infrastructure. This approach, unlike previous studies, allowed participants to directly engage with the search engine and submit their own queries to get their desired search results. \nOur research revealed that search engine results have a substantial impact on individuals, both in terms of positive and negative effects. Significantly, the study participants made more incorrect decisions when they were engaged with topics with unhelpful treatments. Furthermore, it was discovered that there existed a positive correlation between the participants’ level of prior knowledge of health issues and treatments, and their performance in making decisions. One might hypothesize that the results of Pogacar et al. (2017) were due in part of the use of static search result pages rather than a fully interactive search engine, but in our study we found that, even though the participants used a fully interactive search engine, interaction alone was not sufficient for participants to avoid being negatively influenced by the search engine on some search topics.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".