Big problems with ChatGPT

ChatGPT is a powerful new artificial intelligence chatbot, but many people have pointed out that it has some serious flaws. ChatGPT has made more than one mistake.

1. ChatGPT generates wrong answers

Unlike other AI assistants like Siri, ChatGPT does not use the Internet to find answers. Instead, it builds a sentence word by word and selects the next most likely "token" based on its training. In other words, ChatGPT arrives at an answer through a series of guesses, which is one of the reasons it can treat the wrong answer as completely correct.

While it is good at explaining complex concepts, making it a powerful learning tool, it is important not to believe everything it says. ChatGPT is not always correct - at least not yet.

2. ChatGPT has bias built into its system

ChatGPT has been trained to write collectively about past and present human beings around the world. Unfortunately, this means that the same biases that exist in the real world can also appear in the model.

ChatGPT has been shown to produce some terrible answers that discriminate against gender, race, and minority groups, and OpenAI is trying to mitigate this discrimination.

One way to explain the problem is to point to the data as the problem, blaming humans for the biases embedded on the Internet and elsewhere. But part of the blame also lies with OpenAI, whose researchers and developers select the data used to train ChatGPT.

Once again, OpenAI knows this is a problem, and says they are addressing what they call "biased behavior" by collecting feedback from users and encouraging them to flag bad ChatGPT output.

3. ChatGPT may take jobs from humans

After the rapid development and deployment of ChatGPT, the dust has not yet settled and the underlying technology behind it has been stitched into many commercial applications. Among the apps that have integrated GPT-4 are Duolingo and Khan Academy.

The former is a language learning app and the latter is a diverse educational learning tool. Both offer what is essentially an AI tutor, either in the form of an AI character that you can talk to in the language you are learning. Or as AI mentors who can provide tailored feedback on your learning.

On the one hand, this could change the way we learn, potentially making education more accessible and the learning process a little easier. But on the downside, it takes away jobs that have long been held by humans.

Technological advances always lead to job losses, but the speed at which AI is advancing means that multiple industries are facing the same problem. From education to illustration to customer service roles, ChatGPT and its underlying technology will completely reshape our modern world.

4. ChatGPT impacts the education industry

You can ask ChatGPT to proofread your text or point out how to improve a paragraph. Or, you can take yourself completely out of the equation and let ChatGPT do all the writing work for you.

Teachers have tried providing English assignments to ChatGPT and have received better answers than many of their students could do. From writing cover letters to describing the main themes of famous works of literature, ChatGPT can do it all without hesitation.

When students start using ChatGPT to help write essays, schools will have to come up with answers quickly. The rapid deployment of artificial intelligence in recent years is sure to rock many industries, and education is just one of them.

5. ChatGPT could be harmful to the real world

The speed at which natural-sounding text is generated makes it easy for scammers to impersonate people you know on social media. Likewise, discovering phishing emails designed to extract sensitive details from you has been simplified. An added benefit is that ChatGPT can generate text without grammatical errors - once a clear red flag.

Dissemination of false information is also a serious concern, and the scale of the text ChatGPT can generate, combined with the ability to make misinformation sound convincing, is sure to make information on the Internet even more suspect.

6. OpenAI holds all the power

With great power comes great responsibility, and OpenAI has tremendous power. It is one of the first AI companies to truly change the world with multiple generative AI models, including Dall-E 2, GPT-3, and GPT-4.

OpenAI chooses which data to use to train ChatGPT, but this information is not available to the public. We simply do not know the details of how ChatGPT is trained, what data is used, where the data comes from, or what the detailed architecture of the system looks like.

Ultimately, we must blindly trust OpenAI to responsibly research, develop, and use ChatGPT, and OpenAI will continue to develop ChatGPT according to its own goals and ethical standards whether we agree with the methods or not.

Conclusion

There are many exciting things about ChatGPT, but beyond its immediate uses, there are serious problems that deserve to be understood. OpenAI acknowledges that ChatGPT can produce harmful and biased answers, and hopes to mitigate this problem by gathering feedback from users. But even if this were not the case, it generates compelling text that could easily be exploited by people with malicious intent.

Using a brand new technology, it is difficult to predict what problems will arise in the future. So while it may be fun to use, make sure you don't believe everything ChatGPT says.