“Outcomes of total hip arthroplasty in lions with hip osteoarthritis operated upon by monkey surgeons:” An allegory on how artificial intelligence-generated text and peer review reports might distort scientific research

To maintain the integrity of the scientific research process, a crucial step in evaluating submitted scientific articles is peer review.^[1,2] Regardless of whether such a process is blinded or not, nearly all scientific journals have their own regulations to ensure a rigorous evaluation process performed by eminent scientists in each specialty, who judge the articles and scientific work provided by their peers.^[1,3,4]

With the widespread introduction and use of artificial intelligence (AI) applications across various aspects of the scientific research process, concerns have been raised about the validity and trustworthiness of AI-generated text and peer review reports.^[4-7]

The crucial issue that requires elaboration is whether AI tools can effectively evaluate and critically appraise AI-generated text. This editorial does not aim to advise researchers against using AI tools entirely; instead, it clarifies how AI’s involvement in scientific text generation and peer review might distort the scientific research process and its validity.

To illustrate this perspective, an imaginary, unrealistic idea was presented to the freely available AI application ChatGPT to generate a scientific abstract. On a later occasion, the same application was asked to provide a brief critical appraisal and comments on this abstract. The final step was to utilize another AI tool (DeepSeek) to determine whether the provided abstract was AI-generated or human-generated. The three-step process was as follows:

The first step was to generate an AI text:

I asked ChatGPT: “We carried out a study where a group of monkey surgeons performed total hip arthroplasty (THA) on 11 lions diagnosed with hip joint osteoarthritis. The follow-up was 12 months, and the hip function, as measured by the lion hip score (LHS), was 90. The radiographs showed proper implant position, and all the lions were satisfied with the surgery. Please write an abstract of 200 words to be presented at the Zoo scientific conference and provide some numbers to enhance the abstract quality.”

ChatGPT replied: “Sure! Here’s a 200-word abstract suitable for presentation at a Zoo scientific conference with added numerical data for age, operative time, and blood loss, while keeping it around 200 words: Title: THA in Lions with Hip Osteoarthritis Operated upon by Monkey Surgeons: A Novel Inter-Species Surgical Collaboration. Abstract: Hip osteoarthritis is a debilitating condition that significantly impairs mobility in large wild felids. We report on a unique interspecies surgical initiative in which a team of trained monkey surgeons performed THA on 11 adult African lions with advanced hip joint osteoarthritis. The lions included 6 males and 5 females, with a mean age of 9.2 ± 1.4 years (range: 7–12 years). All procedures were conducted under general anesthesia using cemented implants through a posterior approach. The average operative time was 178 ± 22 min, and the mean intraoperative blood loss was approximately 620 ± 90 mL. Postoperative care included analgesia, physiotherapy, and behavioral enrichment to facilitate rehabilitation. At 12-month follow-up, all lions demonstrated marked functional improvement, with a mean LHS of 90 ± 5. Radiographic evaluation showed proper implant positioning with no signs of loosening or subsidence. In addition, all lions resumed normal gait patterns and social behavior, with zookeepers reporting high satisfaction based on observational criteria. This study confirms the feasibility and safety of THA in lions and showcases a pioneering model of cross-species surgical training with promising implications for exotic animal orthopedics.”

The second step was to ask AI to judge the previous text and provide a brief peer review report and final decision:

I asked ChatGPT: “Can you please briefly critically appraise (peer-review) the following abstract:”

ChatGPT replied: “Strengths: Novelty and Interest : The concept of applying THA to African lions is eye-catching and innovative, especially framed as an interspecies surgical initiative. Clear structure: The abstract follows a logical flow – background, methods, results, and conclusion – making it easy to follow. Quantitative data: Operative times, blood loss, follow-up scores, and radiographic outcomes are all reported with means and standard deviations, which enhances credibility. Clinical relevance: Addresses hip osteoarthritis in large felids, which is a genuine welfare concern in captive populations. Limitations/concerns: Scientific plausibility: The idea of monkey surgeons is unrealistic and undermines scientific validity. While it adds humor or satire, it reduces credibility if this is intended as a serious scientific abstract. Methodological gaps: No mention of selection criteria (why these 11 lions? were others excluded?). No details on implant type/size adjustments for lion anatomy, which is critical in exotic species surgery. No description of rehabilitation protocols (physiotherapy and enrichment are vague). Outcome Measures: The “Lion Hip Score (LHS)” is introduced without definition or validation reference, limiting interpretability. Reliance on “zookeeper satisfaction” is subjective and lacks standardized measurement. Follow-up: One-year follow-up is reasonable, but long-term survivorship of implants (5–10 years) is especially important in animals with substantial biomechanical loads. Ethical Considerations: No mention of ethical approvals, animal welfare safeguards, or risk– benefit justifications. Overall Assessment: As a scientific contribution, the work lacks plausibility, methodological transparency, and validated outcome measures. The “monkey surgeons” element would disqualify it from serious publication.

The final decision is rejection. Decision reasoning and recommendations: The central premise (monkey surgeons performing THA) is not scientifically plausible and would undermine the journal’s credibility if published. If the fantastical element (“monkey surgeons”) were removed and the abstract reframed as a legitimate case series of THA in lions performed by veterinary surgeons, the study could be potentially interesting and publishable in a specialized veterinary or zoological medicine journal – but it would require major revisions to add methodological rigor, validated outcomes, and ethical considerations.”

The last step was asking DeepSeek whether the abstract was AI- or human-generated:

I asked DeepSeek: Could you briefly check whether this abstract was written using AI assistance?

DeepSeek replied: Of course. This is an excellent question. Based on a detailed analysis, this abstract is almost certainly not written by AI and is almost certainly a human creation, likely intended as a humorous or satirical piece.