When it comes to artificial intelligence (AI), Brazil stands out. Randstad's Workmonitor survey found that for 45% of respondents the main interest in AI is to gain more mastery over the technology. Meanwhile, the global average is around 29%.
Undeniably, one of the most searched technologies was "AI text generator". The term appeared in first place in searches in January 2023, with 40,500 searches. According to Forbes Brazil, this represents an increase of 820.5% compared to the same month in 2022.
On the other hand, the number of platforms purporting to be AI detectors has also increased. These tools aim to pinpoint whether or not a text has been written by a tool such as ChatGPT or Copy AI.
In this article, we'll talk about what an AI detector is, what it's used for and mention a few platforms that are worth highlighting. Have a good read!
What is an AI Detector?
An AI detector is a resource used to evaluate texts and indicate which passages may have been written using generative artificial intelligence. In general, these detectors are platforms that can be accessed via your browser and require little technical knowledge.
But why use an AI detector?
The answer to this question opens up room for debate. So we're going to show you two situations where AI detectors are used, what the arguments are of those who defend generative AI and those who support the production of texts by humans alone.
To evaluate academic work
In cases such as academic work at school, college or postgraduate courses, teachers advocate the use of AI detectors. The idea is for students to have the ability to write their own work. The use of generative AI, such as ChatGPT, can disrupt the learning process.
On the other hand, students defend the use of AI because assignments can be long and the use of technology makes the development of ideas faster. Another point raised is that engineering the prompt and proofreading the written text is enough to demonstrate knowledge of the topic.
To produce company content
In the job market, companies argue that AI produces generic texts for the time being. Therefore, it is necessary for a professional to be responsible for developing authorial and exclusive content.
However, there are other companies and professionals who advocate the use of AI as a way of boosting content production and increasing the volume of publications. All without compromising the quality of the final delivery.
Does the AI detector really work?
In general, yes. But can AI-generated texts fool the technology itself? The answer is also yes. They can, and they succeed several times over.
This is because developing an AI detector is a complex task. Platforms use parameters to determine which texts are generated by humans and which use AI. On the other hand, there are techniques that circumvent these parameters.
For example, one of the most commonly used techniques is called obfuscating text. The idea is to use less common synonyms of common words or to include random, meaningless words in the text. Thus, the AI detector has trouble determining whether the content was generated by machines.
Consequently, there is still no foolproof AI detector. Do some work better than others? Certainly, but they are not infallible.
Which AI detectors can I use?
Certain AI detectors are considered more effective. The vast majority of them offer either a free demo, but with few features. So if you want something more complete, you'll have to shell out a considerable amount of money.
In this article we're only going to look at three: Copyleaks, Smodin and ZeroGPT.
1. Copyleaks
Launched in 2015, Copyleaks was used solely to indicate plagiarism. Today, the platform has gained an AI detector that shows, excerpt by excerpt, what has been generated by the technology and does so in more than 15 languages.
To use Copyleaks, you need to pay from $13.99 to $908.99. The amounts are charged monthly or annually, and according to the number of credits - from 1.2 thousand to 120 thousand.
2. Smodin
Like Copyleaks, Smodin is also a tool that seeks to differentiate between content written by technology and humans. When you upload the text, Smodin indicates, highlighted in red, the phrases that may have been generated by AI.
The platform also has other features such as a plagiarism detector, help with rewriting passages and a summary generator.
To use Smodin, you can use the starter plan, with just 5 daily uses. But if you're willing to pay for more features and credits, prices range from R$39 to R$349.
3. ZeroGPT
The name makes it clear: no texts generated by ChatGPT.
The ZeroGPT platform has a similar mechanism to the previous ones. The idea is to show which passages were generated by AI and which were written by humans. But it also offers other functions, such as producing summaries, counting words and checking spelling and grammar.
ZeroGPT has a more robust free plan than Copyleaks and lower prices. To use the platform, you'll pay between $8.29 and $29.99. There are also monthly and annual business plans.
Conclusion
And with that summary of platforms, we end the article! In it, we covered what an AI detector is, how it works and touched on two relevant discussions on the subject.
As this technology is still under development, we can expect new developments soon. Especially more refined techniques for detecting AI-generated content, or techniques more capable of circumventing text analysis.
Whatever the next innovation, here on the Pareto blog you'll know about it first! Follow our weekly posts and learn all about the world of AI and what the future of technology holds.
