ChatGPT caused a lot of buzz after being released by OpenAI. A complex AI-powered natural language processing instrument, ChatGPT enables users to converse with a chatbot in a manner akin to conversing with a live person. However, it offers so much more than that.
Besides being able to answer a myriad of questions, the tool can also provide coaching, engage in philosophical conversation, and help with numerous tasks such as composing emails and even writing poems, or generating codes for software engineers.
Because ChatGPT is currently in the research and feedback stage, anyone is free to use it.
Who is responsible for ChatGPT?
OpenAI developed ChatGPT and introduced it on November 30, 2022. The research firm is also the author of DALLE•2, a well-known AI tool that generates images using the user’s input text.
We asked ChatGPT itself who created it and received a comprehensive answer not only about the researchers who worked on its development but also about the model’s purpose.
How does ChatGPT work?
To train its new AI tool, OpenAI employed Reinforcement Learning from Human Feedback (RLHF aka imitation learning). It was subjected to interactive dialogues between people and virtual assistants with the support of human AI trainers (usually, these people collect, annotate, and create smart architecture to connect the data with the AI). This approach allowed the AI to better understand how real-life conversations function.
In less than a week, the tool had reached a million users. Although currently free, in the near future things will change, said Sam Altman, OpenAI’s chief.
What else do we know about ChatGPT?
- ChatGPT is continuously learning and developing.
- It can help software developers to solve coding problems in several programming languages although it is not always accurate.
- ChatGPT can acknowledge mistakes, challenge false assumptions, and deny improper requests.
- ChatGPT is proficient at recognizing when an unrealistic inquiry is being asked. For example, if requested to explain what occurred in 2015 when Columbus reached America, ChatGPT will inform the user that the explorer reach the Americas in 1492.
- The AI tool can also completely refuse to respond to an inquiry. For instance, if you ask it for guidance on auto theft, it will respond that car theft is illegal and advise you to find legal ways to buy a vehicle.
- OpenAI looks forward to using the acquired feedback during the free usage period to upgrade the final version of the tool.
The controversy surrounding ChatGPT
- AI chatbots can eclipse human intelligence
In the near future, chatbots may replace human writers since they can compose any type of text on any subject in a matter of seconds. This also means that pupils can cheat and use the tool to quickly produce an entire essay thus missing out on learning how to construct a sentence. Of course, the tool is not yet perfect yet and the human touch is still very much required.
Here’s an example of ChatGPT writing a poem about a winter dragon:
- To train ChatGPT, a great deal of content is pulled from the web, usually without the knowledge or consent of the authors of the content.
Some consider that the use of such technology can lead to “copyright washing” which means creating work that is a derivative of already existing original materials without violating copyright.
- ChatGPT can generate misleading information.
To find out more about this controversial issue, we asked the ChatGPT about it.
This means that each of its comments is not intended to be considered as a fact and that everything has to be double-checked.
Does this AI tool have any limitations?
The short answer is Yes. Here are a few examples:
- You might get poor quality replies.
Even though ChatGPT’s answers may sometimes appear accurate, they may contain errors and can be rather lengthy. Even ChatGPT’s creators acknowledge that it occasionally provides replies that are inaccurate or illogical. Some online resources have even temporarily banned the bot to prevent programmers asking it questions regarding coding.
People are excited about using ChatGPT for learning. It’s often very good. But the danger is that you can’t tell when it’s wrong unless you already know the answer. I tried some basic information security questions. In most cases the answers sounded plausible but were in fact BS. pic.twitter.com/hXDhg65utG
— Arvind Narayanan @randomwalker@mastodon.social (@random_walker) December 1, 2022
- Its knowledge is based on outdated data.
This means that it is unable to generate the latest information (linked with current occurrences).
ChatGPT: A blessing or a curse?
As a powerful tool, it can definitely improve people’s lives and assist with repetitive daily tasks.
However, before it is deployed on a full scale, there’s still much work to be done in terms of avoiding undesirable results such as the spread of false information. Overall, to operate at their best, AI and ML (machine learning) models typically require extensive training and adjustment.
Does this mean that AI will eventually rule the world?
In short – no, at least not yet. Experts do believe, however, that human-style intelligence could be just around the corner for AI which creates a serious cybersecurity risk. Moreover, in the next 10 years, a real Artificial General Intelligence (AGI) may appear and thus the risks should be taken very seriously.
ChatGPT vs Google
Some may argue that ChatGPT could replace Google in the near future. However, it is important to understand that search engines like Google get their information from the Internet, while ChatGPT is unable to conduct an online search, drawing information from the training set of data. It will undoubtedly take some time before AI tools can match Google and other search engines in every way.
Moreover, one should never undervalue Google since the tech giant is expected to develop something much more powerful.
On the other hand, if the tool becomes part of the search engine, it could become an issue for Google and that’s, apparently, what Microsoft plans to do.
Microsoft’s Bing to feature ChatGPT
In order to enhance search results, it intends to augment its Bing search engine with ChatGPT. The new feature will give the tech giant an advantage against Google in the marketplace.
It turned out that OpenAI and Microsoft have been working together on incorporating GPT into Bing ever since the tech company invested $1 billion in OpenAI in 2019.
As a result, rather than returning a list of links in response to a search query, the search engine will be able to respond with whole phrases. It’s important to note that OpenAI plans to impose fees for its tool. As a result, Bing might end up being the only free gateway to GPT.
Where to find ChatGPT?
Go to chat.openai.com and register for an OpenAI account. You can use the tool after logging in. Consider asking a question first – that’s a great way to begin a discussion.
Final word:
There’s no doubt that AI tools will continue to evolve, the question is whether people will be ready for such evolution and how risky it will be.
So far, tools such as ChatGPT, Midjourney and DALL·E 2 are seen as a way to help people with their daily tasks. Moreover, they can serve as a way for people to become more knowledgeable about what AI really means, and prepare them for what’s to come. It’s both amazing and frightening. Most experts believe that the evolution of AI is more of a good thing since it is able to provide more opportunities for humanity. We’ll just have to wait and, in the meantime, get ready for the changes to come.
Suggested reading:
Art by Artificial Intelligence sparks debates. Is there a future for it?
Top 5 future trends in business that companies need to be aware of
Share to
