OpenAI, the world’s leading artificial intelligence laboratory, has recently launched its new open-source classifier to detect potential misuse or abuse of synthetically generated content. This exciting and innovative tool is designed to detect if any AI technology has been used to create realistic articles or videos that malicious actors could misuse.
OpenAI hopes this technical breakthrough will provide much-needed protection from unethical applications of synthetic media being used in deceptive activities such as deepfakes and fake news, helping people make more informed decisions about the information they consume in our digital world.
OpenAI has taken the initiative to make AI detector tools obsolete by introducing an AI Text Classifier that can differentiate between text written by humans and text generated by AIs from multiple sources. Yesterday, they announced the launch of this classifier.
OpenAI stated that although it is difficult to detect all AI-generated text accurately, they are confident that classifiers can be utilized to recognize and combat deceiving claims made by humans claiming the text was written by AI.
OpenAI has released a new classifier that is more dependable when assessing the text generated by modern AI systems. This has been made available to the public to receive feedback and evaluate if these tools are beneficial. Compared to its earlier classifier, it is much more reliable.
OpenAI has revealed a fine-tuned GPT model known as an AI Text Classifier that estimates the probability of text being created by artificial intelligence from sources such as ChatGPT. This classifier is available to encourage conversations on AI literacy.
Cons Of AI Text Classifier
- A text length of at least 1,000 characters is required for the classifier to operate—about 150–250 words.
- The classifier doesn’t always have a correct result; it can mistake AI-generated texts for those written by humans and vice versa.
- Editing AI-generated material is straightforward and can be employed to influence the categorization process.
- The classifier will probably not be very accurate regarding text written by people who do not speak English and children, as it has mostly been trained on English content created by adults.
- It is impossible to determine something very generic and foreseeable with certainty, such as the fact that the first five letters of the alphabet will be the same.
- OpenAI has noted that neural network-based classifiers tend not to function very well when dealing with data distinct from the material used in training. When presented with inputs substantially dissimilar to the text in our training set, the classifier can sometimes be highly confident of a false prediction.
The AI Classifier was tested using an essay that ChatGPT created, and the outcome is presented here.
ChatGPT Ban
In the wake of numerous schools and educational institutions expressing worries about pupils utilizing ChatGPT for answers, criticism has been leveled against ChatGPT from academic organizations worldwide. It is because it can effortlessly compose essays, codes, scripts, and poetic and literary works of all sorts.
OpenAI has created an introductory guide for educators to better understand the utilization of ChatGPT and its potential constraints. The resource provides information on how it can be used and considerations that should be taken into account due to the criticism received.
OpenAI is collaborating with American educational institutions to devise solutions for the problem.
Despite concerns about the potential misuse of its technology, OpenAI is moving forward with plans to release an AI-generated content classifier to the public. The classifier will be available for free and can be used by anyone who wishes to monitor or control AI-generated content. While it is still early days for this technology, it has the potential to play a significant role in helping to prevent the misuse of AI-generated content.
Source: analyticsindiamag.com
