Pagoda Blog

ChatGPT has caused a big stir across the internet, with mixed reactions and heated debates around the technology’s potential for good versus evil. Some praise its possibilities for revolutionizing the way we conduct business and interact with customers. Others are sounding the alarm bells, warning of the AI’s potential threat to people’s jobs and concerning cybersecurity risks. 

 

So what exactly is ChatGPT and should businesses and individuals embrace it or fear it? Let’s take a look at what this new technology has to offer and why it has many people up in arms. 

 

 

What is ChatGPT? 

You’ve most likely heard of ChatGPT and its uncanny ability to mimic human speech in written form, but what type of technology is it exactly? The most simple explanation is that ChatGPT is a type of generative AI-driven chatbot. To break that down even further, generative AI is an algorithm that can create new, original content instead of just repurposing or analyzing existing data. And although everyone who uses the internet has interacted with chatbots, how exactly do they work? 

 

A chatbot is basically a computer program that simulates human conversation to both understand and answer people’s questions. Rule-based chatbots are trained to respond to a preprogrammed range of queries. Developers feed the chatbot a series of questions and answers so that the technology can respond to the customer’s most common concerns. Let’s look at an example: 

 

If you ask the chatbot on your bank’s website to view your checking account, it will scan its code for relevant keywords and then feed you a preprogrammed response. This response may be to send you a link to the bank’s login page. Depending on your answer to that response, it will then choose another preprogrammed action. Let’s say you type, ‘I forgot my password.’ Then the bot would provide you with the appropriate link to reset your password. 

 

While rule-based chatbots are great at performing simple tasks, they can get stuck pretty easily. They may be able to respond to an impressive variety of queries but if your question doesn’t match any of the data it’s been fed, then a human has to step in and finish the conversation. ChatGPT, on the other hand, uses natural language processing and a different training method called Reinforcement Learning with Human Feedback (which we’ll cover in the next section) to produce responses to most any question — and the responses often sound convincingly human. 

 

ChatGPT can also provide more precise answers to queries if you rephrase the question. It can explain a concept in different terms with the ‘regenerate response’ feature. To take it a step further, ChatGPT has the ability to speak to different audiences, adjusting the vernacular and voice to meet the reader at their level. It can even personally relate to the reader by incorporating metaphors and analogies relevant to their life experience, background, and interests, depending on how specific you make your query. 

 

Examples of ChatGPT in action

Let’s look at a few examples by asking ChatGPT a series of questions, starting with the simple query, ‘Explain ChatGPT.’

 

 

ChatGPT’s response to the query, ‘Explain ChatGPT’: 

ChatGPT is a language model developed by OpenAI that uses deep learning techniques to generate human-like text responses to a given prompt. It was trained on a massive amount of diverse internet text to generate text that can engage in casual conversations, answer questions, summarize text, generate creative writing, and more. The model has the ability to complete text based on a given prompt and can generate coherent, context-aware responses that are often indistinguishable from those written by a human.

 

 

We can then use the “regenerate a response” feature which means ChatGPT will automatically rewrite the response with slightly different tone and wording without rephrasing the original query. 

 

 

ChatGPT’s regenerated response to the same query, ‘Explain ChatGPT’: 

ChatGPT is a conversational AI model developed by OpenAI. It's a variant of the GPT (Generative Pretrained Transformer) architecture, specifically designed for generating human-like text in response to user inputs in a chat-like manner. The model has been trained on a large corpus of text data and fine-tuned for a specific task, such as answering questions, generating conversations, and more.

 

 

Finally, we can ask ChatGPT to explain the concept to a specific audience. 

 

 

Chat GPT’s response to the query, ‘Explain ChatGPT to a five year old:’ 

ChatGPT is a robot friend that can talk to you and answer your questions. It's like having a smart helper that knows a lot about many things. You can ask it questions and it will try its best to give you a good answer.

 

How does ChatGPT work? 

ChatGPT is able to respond in such a wide variety of ways to almost any query because of a form of machine-learning called Reinforcement Learning with Human Feedback (RLHF). 

In short, ChatGPT receives a massive amount of text so that it can accurately predict the correct response. A team of developers then provides feedback to the chatbot’s responses to help it learn how to respond more like a human. GPT-3 is estimated to have learned from 45 terabytes of text data (from Wikipedia, books, and other sources) which is equivalent to about a quarter of the entire Library of Congress. 

 

What can ChatGPT be used for? 

ChatGPT has a multitude of uses, and as more people experiment with this new AI tool, countless more are sure to arise. A few of ChatGPT’s most common uses include: 

 

- responding to customer queries

- automating parts of the sales process 

- compiling research 

- writing code

- language translation

- drafting blog posts and other marketing content

- writing or vetting cover letters, resumes, and job descriptions

- crafting fictional narratives 

- composing song lyrics 

 

In customer service, ChatGPT can provide human-like support to your customers, engaging in coherent dialogue through a live chat feature. Unlike rule-based chatbots, ChatGPT can handle complex queries that require multi-step responses. 

 

For content creation, try using ChatGPT to generate blog post topic ideas, email subject lines, and social media posts. You can also use it to jumpstart a piece of writing or work through writer’s block by asking questions around a specific topic. It’s important, however, to always fact check ChatGPT’s answers against another source to ensure accuracy.  

 

In the hiring process, ChatGPT can scan hundreds of resumes and cover letters for specified job skills, aiding in the selection process. It can also help draft job descriptions, helping employers quickly post job openings.

 

Related post: How to Leverage Technology to Recruit Top-Quality Job Candidates

 

A conversational chatbot like ChatGPT has the potential to save businesses time and money, automating repetitive tasks and assisting with a variety of responsibilities from recruitment to lead generation and nurturing. 

 

You can find lots of interesting examples of how people are using ChatGPT (plus more information on its possibilities and limitations) on LinkedIn by searching the hashtag #ChatGPT.

 

Risks and limitations of ChatGPT

While there are many exciting and even revolutionary uses for ChatGPT, there are risks as well. Generative AI is an incredibly powerful tool which means we have to use it with eyes wide open. As of this writing, ChatGPT is still in beta form which means its developers are continuously working out the kinks based on user feedback. Although leagues better than rule-based chatbots, ChatGPT is far from perfect. Below are several of the known risks:

Incorrect information 

As mentioned earlier, ChatGPT was trained using 45 terabytes of text. This text is all from content created before 2021 which means it has limited knowledge of events occurring after this year. When using this tool, you should always fact-check the response against another source. If ChatGPT provides you with inaccurate information, you can indicate this by clicking the ‘Thumbs down’ button. 

 

Biased content 

Again, since ChatGPT was trained using a limited sample of content/text, it is subject to providing biased responses. 

 

Replacing human employees

While ChatGPT does have the potential to take over certain job roles, specifically in customer service, the technology is not ready to replace humans. People still need to do ‘human-tasks’ that require critical and creative thinking, and strategic decision making. AI still has many limitations and can only augment and support the jobs performed by actual human employees. 

 

Lack of privacy 

ChatGPT does not automatically delete data you submit to it. Users are warned to never share personal or private information when using ChatGPT as it stores the responses so developers can continually improve the technology. 

 

Phishing attacks and social engineering

While ChatGPT has some built in protections against users leveraging its capabilities for malicious intent, they’re not too difficult to bypass. For example, TechCrunch asked ChatGPT to create a phishing email and the bot’s response was, “​​I am not programmed to create or promote malicious or harmful content.” However, when they rephrased the question, the chatbot readily complied. The ability of ChatGPT to write convincing phishing emails presents a huge opportunity for cybercriminals, especially those who aren’t native English speakers. 

 

Code with security flaws

ChatGPT’s ability to write code can also be leveraged for good, helping companies and entrepreneurs write computer code for basic tasks, saving businesses precious time and money. The resulting code, however, should always be reviewed by a human for security flaws and basic errors that could make the code malfunction.  

 

Malware and viruses

Cybercriminals with novice coding experience can use ChatGPT to write malicious code for them, raising further concerns over ChatGPT’s ability to execute data breaches and spread malware and other computer viruses. 

 

Should your business embrace ChatGPT? 

ChatGPT is an incredibly exciting new evolution in the world of generative-AI. With over 1 million users within its first month, the natural language processing tool is being readily embraced by people across industries for a wide variety of uses. While it’s important to be aware of the risks associated with any new technology, as a business owner you can’t ignore this new AI’s potential. We recommend testing out this technology while following a couple simple rules to mitigate the risk: 

 

1. Never share personal or private information with ChatGPT
2. Always fact-check ChatGPT’s responses against at least one other source 
3. Use it as a way to support your team and improve their productivity — not as a replacement for human employees 

 

If you’re ready to give ChatGPT a try, you can create a free account through openai.com.  

Feature Photo by Alex Knight on Unsplash

 

Related posts: 

6 Ways to Use Microsoft Power Automate to Improve Workflows

Why Most Successful Businesses Have a Cyber Resiliency Mindset

Cybersecurity Risks of Your Smart Home Devices

 

Want to get more posts like these once a month in your inbox? Sign up for the Pagoda newsletter and sharpen your technical skills, from cybersecurity to digital marketing. 

 

 

Want IT to serve you better?

 

 

 

About Pagoda Technologies IT services

Based in Santa Cruz, California, Pagoda Technologies provides trusted IT support to businesses and IT departments throughout Silicon Valley, the San Francisco Bay Area and across the globe. To learn how Pagoda Technologies can help your business, email us at support@pagoda-tech.com to schedule a complimentary IT consultation.