ChatGPT is a chatbot released by OpenAI. Since its release in November 2022, the tool which is based on the OpenAI GPT-3 family of large language models has been in the buzz since then.
The chat tool has been phenomenal, having passed several tests and being able to write essays, poems, spectacular messages, and copy among others.
However, the chatbot isn’t all-knowing. It has limitations, it isn’t human, so, it has shortcomings, even human beings do have shortcomings.
In one of our articles, we pointed out that ChatGPT is unable to answer certain types of questions since it relies on the input it receives.
Table of Contents
Despite the sterling performance of the chatbot tool, here are the tasks ChatGPT is not capable of doing:
Physical Tasks:
ChatGPT is software and there is no physical way to interact with the world. The tool has no hands nor legs, it cannot lift objects.
Unlike human beings, ChatGPT can’t move around or perform any other physical activity.
Unlike a robot which can engage in physical activities like lifting objects, chatgpt can not.
It is purely a language model that has been trained to respond to questions based on the input it is trained on.
It is designed to process text data and generate text output. It is not programmed to control physical objects or perform physical actions.
So, carrying out physical task is one of the inabilities of ChatGPT. It is trained to understand and generate natural language, which is the primary way humans communicate.
Inability to protect user privacy
ChatGPT is a model that runs on the server and has no way to protect user privacy, cannot encrypt or decrypt input or output.
Unable to access the internet:
ChatGPT only relied on the information it is trained on, its output is based on the info that is fed into it.
It has no Internet access and cannot search for information or access external resources beyond what it has been trained on.
Just like I was initially curious, you might want to ask: where does ChatGPT get the output if it says it has no internet access?
The interesting thing is that the language model tool has been trained on a large text dataset that contains a wide range of information and knowledge.
This dataset was used to train the model so that it could generate text based on the input it received.
The model uses this dataset to generate text by looking for patterns and relationships in the data and using that information to generate new text.
Note that although the model was trained on a large dataset, this dataset is limited to the information that was available when the model was trained.
The model does not have access to the internet, so it cannot search for new information or access external resources beyond what it has been trained on.
That’s why it is still unable to answer queries that are related to a developing story or event.
Another shortcoming of ChatGPT is that it cannot guarantee the accuracy, reliability, or validity of the output.
It generates its output based on patterns it has learned from the data it has been trained on, it may not reflect the current state of the world or the information available on the internet.
No common sense of understanding:
ChatGPT is trained on a large text dataset, but it does not have a deep understanding of the world like a human and does not have common sense.
It cannot react to situations like humans because it cannot understand the context of the conversation and cannot react as a human would.
Inability to understand other forms of data:
It is designed for word processing, as such, it may not be able to understand or respond to other forms of data such as images, audio or video.
If the buzz around the tool made you believe that it can handle all questions and wonder why can’t chatGpt understand other forms of data besides text?
It is a language model, trained on a large text dataset and uses this dataset to generate text based on the input it receives.
The model uses advanced techniques such as machine learning and neural networks to understand and generate text.
The techniques it has been trained on don’t include images, sound or video. Therefore, the model may not be able to understand or respond to other forms of data because other forms of data, such as images or videos, requires specialized models and techniques, such as computer vision for images and natural language processing for videos.
Understanding other forms of data is an active area of research and there are models like DALL-E, DALL-E 2 that can generate images from natural language descriptions.
Inability to perform a real-time task:
ChatGPT is unable to perform tasks that require real-time processing, such as voice recognition, image processing, and other tasks that require high accuracy and speed.
Tell the chatbot to tell you January 30, 2023, real-time proceedings of the US Congress, it can’t.
Inability to make decision
The chatbot is not capable of making decisions on its own, it depends on input from humans.
It can only generate text based on given inputs and has no reasoning or critical thinking ability.
But reviews from millions of ChatGPT say the chatbot makes suggestions, why can’t it make decisions?
Relying on the dataset to generate text based on the input it receives, it can generate text that may make suggestions or recommendations but has no decision-making capability.
Making a decision requires a certain level of understanding and consideration of the situation, the options available, and the potential outcomes.
It also requires the ability to weigh the pros and cons of different options and choose the best one.
ChatGPT can generate text that can make suggestions or recommendations based on patterns it has learned from the dataset it has been trained on, but it has no way of understanding the situation, justifying options, or weighing up the pros and cons.
It is also important to note that decision-making is often a complex process that involves many factors including ethics, laws and regulations, which the model is not programmed to take into account.
Even a person with knowledge and experience in a particular field can have difficulties in making a reasonable decision
Multi-language limitations
ChatGPT is trained in English, it does not understand or respond to other languages.
ChatGPT is a language model that has been trained on a large dataset of English text, so it is primarily intended for understanding and generating English text.
However, it can generate text in other languages to some extent, including Chinese.
ChatGPT’s ability to understand and generate Chinese text depends on the quality and quantity of Chinese text data on which the model has been trained.
The more Chinese text data the model has been trained, the better it will be able to understand and generate Chinese text.
It should be noted that understanding and generating text in a language such as Chinese, which has a different writing system and grammar rules than English, is a complex task that requires specialized techniques and models.
In addition, the model would need to be tuned and trained on Chinese and other language text data to improve its performance.
Conclusion
In conclusion, ChatGPT is an advanced language model that can generate text based on input data, but is unable to perform physical tasks, access the Internet, understand other forms of data, make decisions, complete tasks in real-time, have common sense, protect user privacy or supporting multiple languages.
In addition, the lack of access to external information may make it less useful in certain situations where up-to-date information is needed.
ChatGPT is a machine learning model, so it makes mistakes and produces output that is not always accurate or appropriate.
As a result of the limitations of ChatGPT, users need to verify the output generated by the chatbot.