Model Card for ai-text-detector-model
Model Details
Model Description
This model is a fine-tuned DistilBERT sequence classification model to detect whether a given text is AI-generated (e.g., by ChatGPT, GPT-2/3) or Human-written.
It was trained on a combination of AI-generated texts and human-authored content.
- Developed by: Ahmed Iqbal
- Funded by [optional]: Self-funded
- Shared by: Ahmed Iqbal
- Model type: Transformer-based binary classifier (DistilBERT)
- Language(s) (NLP): English
- License: MIT (you may change this to Apache 2.0 if preferred)
- Finetuned from model:
distilbert-base-uncased
Model Sources
- Repository: https://huggingface.co/ahmediqbal/ai-text-detector-model
- Demo [optional]: Can be easily built using Hugging Face Inference API or Gradio
Uses
Direct Use
- Detect whether a text is AI-generated or Human-written.
- Useful in applications like plagiarism detection, content moderation, or authenticity checking.
Downstream Use
- Can be integrated into web apps for AI content detection.
- Can be further fine-tuned with domain-specific data (e.g., academic writing, creative writing).
Out-of-Scope Use
- Should not be used for high-stakes scenarios (e.g., exams, hiring, legal decisions).
- May not generalize well to languages other than English.
- Not reliable for adversarially modified text (e.g., humanized AI text).
Bias, Risks, and Limitations
- Bias: Model may misclassify some human-written texts that resemble AI style.
- Risks: Over-reliance on automated detection may lead to false accusations.
- Limitations: Works best on English text only. Accuracy may decrease for very long or domain-specific texts.
Recommendations
- Always use this model as supportive evidence, not as a sole decision-maker.
- Combine with human review in critical cases.
How to Get Started with the Model
from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
model_id = "ahmediqbal/ai-text-detector-model"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(model_id)
classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)
text = "This is a sample sentence."
print(classifier(text))