You may have heard a lot of buzz about ChatGPT. However, after talking to my students and parents, I felt that not everyone has a solid foundation of what ChatGPT is, what it can do, and what it means for the future. So, I spent many hours digging into it and wanted to share the good, the bad, and the ugly about ChatGPT.
ChatGPT is a state-of-the-art chatbot, an Artificial Intelligence (AI) model developed by OpenAI. It belongs to a category called 'Generative AI,' a type of algorithm that can create text, images, or other data using AI. First and foremost, it is an intelligent bot. So, it not only answers your questions like Alexa, Google Home, and Siri do, it takes the conversation to the next level. The bot remembers everything you said and can have an intelligent discussion with you on any topic, just as your fellow humans do, but with a twist – it knows almost everything about almost every field.
Tactically speaking, it can create human-like text and essays on many topics, answer questions, translate information, classify, and summarize. Its incredible power comes from (1) the data it is trained on and (2) the AI model itself. In this case, ChatGPT was trained on data from internet websites, books, articles, and other public sources. Its input data was preprocessed and filtered to remove low-quality data and ensure data comes from diverse and well-balanced sources. The AI model itself is a highly advanced one. It uses something called transformer architecture (T in GPT), a type of neural network. A neural network is a system of layers of interconnected "neurons" that process and transmits information (just like those found in a human brain). In addition, the ChatGPT uses unsupervised learning, where the bot learns by analyzing patterns and structures in the trained data without any manual intervention (supervision).
The ChatGPT was made available freely in November. It ramped up to an estimated 100 million active monthly users by Jan 20223, just within two months, surpassing the speed with which TikTok (took nine months to 100 million users), Instagram (2.5 years to 100 million users), and other consumer applications got to that magic 100 million users.
The fantastic thing about ChatGPT is that it learns from each interaction. In other words, it improves each time you ask a question (continuous improvement through feedback). So as more and more people on the internet use it, it gets exponentially better and more intelligent. No wonder it inspired awe and fear in education and other industries. The ChatGPT is so good at creating articles that resemble those made by humans that it can easily evade the plagiarism software many institutes use to check for cheating. So, NY City public schools have blocked access to the ChatGPT AI model on their networks and devices, fearing that the bot can be used to cheat on assignments and essays.
Luckily, a Princeton undergrad recently created GPTZero, which can scan writing to determine whether it was created by a human or by an AI bot such as ChatGPT. However, the cat-and-mouse game will go on where ChatGPT gets better at pushing the envelope of what is human-generated writing vs. what's bot-generated.
The Good:
At a fundamental level, ChatGPT is just like any other tool. Tools such as programmable calculators, home PCs, GPS, IPADs, Alexa, and Siri have fundamentally changed how we work, live, and interact with others. We can expect the same with ChatGPT. It is a tool we can leverage to improve our success in many things we do at work and home. Parents should engage their kids in various aspects of the Chat Bot. If possible, use the tool along with the child. I did that with my 13-year-old kid, and we had a great time using ChatGPT. At the same, I had a frank conversation about where it is allowed and where it is not allowed.
See the response below given by the bot when asked about the best way kids can leverage ChatGPT:
ChatGPT can be used for several educational and entertaining purposes for kids, including:
Learning and education: ChatGPT can answer questions and provide information on a wide range of topics, making it a great tool for kids to learn about science, history, and more.
Creative writing: ChatGPT can be used as a tool to help kids generate ideas and get inspired for their own writing.
Language learning: ChatGPT can be used to practice and improve language skills by conversing in different languages.
Fun and games: ChatGPT can also be used for fun activities like playing trivia games or answering silly questions.
It's important to supervise and guide kids when they use ChatGPT or similar technologies, and to help them understand the limitations and biases of AI models.
The Bad:
Using ChatGPT will not build critical thinking or problem-solving, or teamwork skills. Some of the answers may be misleading or confusing. Like TikTok or YouTube shorts, it can waste a child's time and take her away from schoolwork or friends. The bot may not fully get the context of your kid's education needs. It may even lack the context of your kid's question. The bot will not help kids build essential skills such as creativity, diligence, and discovery . These skills are essential for academic and life success. See the response below given by the bot when asked about its limitations:
As an AI language model, I have several limitations, including:
Lack of context: I can only respond to questions based on the information provided to me, and I don't have access to the world's context or external information unless you provide it.
Bias: I was trained on a diverse range of text data, but this data may contain biases, inaccuracies, or outdated information.
Understanding complexity: I may struggle to understand complex or abstract concepts, especially those that require real-world experience.
Creativity: While I can generate text that is grammatically correct, I don't have the capacity to truly understand emotions, humor, or original ideas.
Knowledge cut-off: I was trained on text up to 2021, so I may not have information on recent events or developments.
The Ugly:
Some bad people are using ChatGPT to create malware, posing security threats to many institutions or putting individuals at risk of getting hacked. For example, Check Point company, a leading provider of security software to governments and corporations, recently published an article noting that 'that code generation can help less-skilled threat actors effortlessly launch cyberattacks.' In short, ChatGPT can make it easy for cybercriminals to create code and use it for malicious purposes such as stealing credit card numbers, proprietary content or conduct other fraudulent activity. When asked about how cybercriminals can abuse ChatGPT, the bot replied as below:
Cybercriminals can abuse AI language models like ChatGPT in several ways, including:
Phishing scams: ChatGPT or similar models could be used to impersonate a trusted entity, such as a bank or a government agency, and trick people into revealing sensitive information.
Spreading misinformation: Cybercriminals could use ChatGPT to generate misleading or false information to spread on social media and manipulate public opinion.
Generating fake reviews or testimonials: ChatGPT could be used to generate fake positive reviews for products or services to mislead potential customers.
Automated cyber attacks: ChatGPT or similar models could be trained to generate code or scripts for automated cyber attacks.
It is important to be cautious and verify information from sources you trust before acting on any information provided by AI models like ChatGPT.
In summary, ChatGPT is an exciting tool for specific purposes:
Parents and Kids can benefit from learning how to use it.
When used correctly, ChatGPT complements existing skills and learning.
The tool signals the beginning of a plethora of other intelligent agents and robots we will see (so, what jobs will be relevant in the future, the topic of my next blog).
Please feel free to share your ideas and thoughts below
Comments