Can we spot fake public comments generated by ChatGPT(-3.5, -4)?: Japanese stylometric analysis expose emulation created by one-shot learning.

Public comments are an important opinion for civic when the government establishes rules. However, recent AI can easily generate large quantities of disinformation, including fake public comments. We attempted to distinguish between human public comments and ChatGPT-generated public comments (including ChatGPT emulated that of humans) using Japanese stylometric analysis. Study 1 conducted multidimensional scaling (MDS) to compare 500 texts of five classes: Human public comments, GPT-3.5 and GPT-4 generated public comments only by presenting the titles of human public comments (i.e., zero-shot learning, GPTzero), GPT-3.5 and GPT-4 emulated by presenting sentences of human public comments and instructing to emulate that (i.e., one-shot learning, GPTone). The MDS results showed that the Japanese stylometric features of the public comments were completely different from those of the GPTzero-generated texts. Moreover, GPTone-generated public comments were closer to those of humans than those generated by GPTzero. In Study 2, the performance levels of the random forest (RF) classifier for distinguishing three classes (human, GPTzero, and GPTone texts). RF classifiers showed the best precision for the human public comments of approximately 90%, and the best precision for the fake public comments generated by GPT (GPTzero and GPTone) was 99.5% by focusing on integrated next writing style features: phrase patterns, parts-of-speech (POS) bigram and trigram, and function words. Therefore, the current study concluded that we could discriminate between GPT-generated fake public comments and those written by humans at the present time.

Distinguishing ChatGPT(-3.5, -4)-generated and human-written papers through Japanese stylometric analysis.

In the first half of 2023, text-generative artificial intelligence (AI), including ChatGPT from OpenAI, has attracted considerable attention worldwide. In this study, first, we compared Japanese stylometric features of texts generated by ChatGPT, equipped with GPT-3.5 and GPT-4, and those written by humans. In this work, we performed multi-dimensional scaling (MDS) to confirm the distributions of 216 texts of three classes (72 academic papers written by 36 single authors, 72 texts generated by GPT-3.5, and 72 texts generated by GPT-4 on the basis of the titles of the aforementioned papers) focusing on the following stylometric features: (1) bigrams of parts-of-speech, (2) bigram of postpositional particle words, (3) positioning of commas, and (4) rate of function words. MDS revealed distinct distributions at each stylometric feature of GPT (3.5 and 4) and human. Although GPT-4 is more powerful than GPT-3.5 because it has more parameters, both GPT (3.5 and 4) distributions are overlapping. These results indicate that although the number of parameters may increase in the future, GPT-generated texts may not be close to that written by humans in terms of stylometric features. Second, we verified the classification performance of random forest (RF) classifier for two classes (GPT and human) focusing on Japanese stylometric features. This study revealed the high performance of RF in each stylometric feature: The RF classifier focusing on the rate of function words achieved 98.1% accuracy. Furthermore the RF classifier focusing on all stylometric features reached 100% in terms of all performance indexes (accuracy, recall, precision, and F1 score). This study concluded that at this stage we human discriminate ChatGPT from human limited to Japanese language.

A rapid screening method for thermal conductivity properties of thermal insulation materials by a thermochemiluminescence probe.

Acridine-based 1,2-dioxetane as a thermochemiluminescence (TCL) probe for temperature sensing exhibited an excellent response for temperature in the range of 85-130 °C with favorable sensitivity and good resolution. The proposed TCL probe could be applied to screen thermal conductivity properties of different thermal insulation materials.

[Value of percutaneous oxygen partial pressure monitoring in evaluating the prognosis of patients with acute heart failure].

OBJECTIVE: To explore the value of percutaneous oxygen partial pressure monitoring in prognosis evaluation of patients with acute heart failure (AHF). METHODS: A total of 91 patients with AHF due to various reasons admitted to the emergency department of the First Affiliated Hospital of China Medical University from July 2017 to June 2018 were enrolled. Dynamic monitoring data of arterial blood gas, percutaneous oxygen partial pressure monitoring and noninvasive cardiac output monitoring (NICOM) of all of the patients at the time of diagnosis (before treatment) and 6 hours after treatment were recorded, including arterial partial pressure of oxygen (PaO2), arterial partial pressure of carbon dioxide (PaCO2), blood lactic acid (Lac), percutaneous oxygen partial pressure (TcPO2), percutaneous carbon dioxide partial pressure (TcPCO2), cardiac output (CO) and stroke volume (SV). The 10-minute oxygen challenge test value (OCT), oxygen and carbon dioxide offsets were calculated. The patients were divided into survival group and non-survival group according to 28-day survival situation, and the differences in above parameters were compared between the two groups. The receiver operating characteristic (ROC) curve was drawn to analyze the predictive value of percutaneous partial oxygen pressure monitoring for the prognosis of patients with AHF. RESULTS: All the 91 patients were enrolled in the analysis, among whom 26 died on 28 days and 65 survived, with a mortality of 28.6%. Before treatment, 10-minute OCT in the non-survival group was significantly lower than that in the survival group [mmHg (1 mmHg = 0.133 kPa): 41.0±3.9 vs. 45.6±3.2, P < 0.01], and the carbon dioxide offset was significantly higher than that in the survival group [(0.51±0.11)% vs. (0.37±0.11)%, P < 0.01]. However, there was no statistically significant difference in PaO2, PaCO2, TcPO2, TcPCO2, oxygen offset, CO, SV or Lac between the two groups. After 6 hours of treatment, TcPCO2, oxygen offset and carbon dioxide offset in the non-survival group were significantly higher than those in the survival group [TcPCO2 (mmHg): 36.0±2.8 vs. 33.2±2.8, oxygen offset: (0.25±0.05)% vs. (0.22±0.06)%, carbon dioxide offset: (0.29±0.12)% vs. (0.16±0.13)%, all P < 0.05], TcPO2, 10-minute OCT, CO and SV were significantly lower than those in the survival group [TcPCO2 (mmHg): 36.0±2.8 vs. 33.2±2.8, 10-minute OCT (mmHg): 49.1±4.5 vs. 53.6±5.5, CO (L/min): 4.9±0.5 vs. 5.3±0.5, SV (mL): 57.8±3.5 vs. 64.4±4.8, all P < 0.01]. However, there was no statistically significant difference in PaO2, PaCO2 or Lac between the two groups. ROC curve analysis showed that the area under the ROC curve (AUC) of 10-minute OCT onset predicting the 28-day death of patients with AHF was 0.802; when the optimal cut-off value was 43.5 mmHg, the sensitivity and specificity was 77.3% and 68.0%, respectively. The AUC of carbon dioxide offset was 0.812; when the optimal cut-off value was 0.46%, the sensitivity and specificity was 86.4% and 68.0%, respectively. CONCLUSIONS: Percutaneous oxygen partial pressure monitoring can be used as a reliable indicator for prognosis evaluation of patients with AHF. Increased carbon dioxide offset and decreased 10-minute OCT suggest poor prognosis, and the prognosis of patients with AHF should be evaluated by dynamic monitoring.