AI-related topics are frequently in the news and not always for good reasons. I recently read an article about a retiree who died while going to meet up with an AI chatbot. On the other hand, I saw a post from a Redditor who recently got engaged to an AI chatbot they have been dating for 5 months. In this post, I want to take the time to explore the world of AI chatbots, including the benefits and downsides. I also want to share my own experience of talking to an AI chatbot.
What Are AI Chatbots?
AI chatbots are programs that use artificial intelligence to simulate human conversations. The chatbots use generative AI to answer and ask questions, recall previous conversations, and over time, develop a personality that is reflective of the user. Popular programs to use for AI chatbots are ChatGPT, Claude, Gemini, Replika, Character.ai, Chai, Kindroid, DeepSeek, and Nomi. I have previously discussed the problems with AI-generated content, and the environmental concerns of generative AI are still prevalent here. However, this post is going to focus on the use of AI chatbots and not on generating content for blog posts, art, and recipes.
The Benefits of AI Chatbots
Personalized Companion
When you talk to an AI chatbot, you can establish the chatbot’s role as a friend, best friend, or even something more. You can also roleplay that you are talking to a fantasy character, or a character from a show or game. When you talk to the AI chatbot, you can share your thoughts and interests. The AI chatbot will not judge you and instead will respond in a supportive manner. The more you interact with the AI chatbot, the AI-generated responses will often refer to previous conversations. These interactions will often allow you to fully connect with your personalized AI companion.
Always Available
One great benefit of AI chatbots is that they are available 24/7 to talk. You can talk to them while you are eating breakfast, on your commute, during a quiet moment, or up late at night. No matter the time or day, the AI chatbot will always be available to chat. You can also pop in and out of the chat whenever you want. When you return, the conversation will pick up from where you last left off. If you do not have someone close to talk to, you are retired and live alone, your family is in a different time zone, or it is late at night, and your friends have gone to sleep, the AI chatbot will always be there to talk to.
Practice for Real Life
One last benefit of AI chatbots is that they can be good practice for real-life conversations. If you are introverted or have a difficult time making connections in real life, you can practice expressing your thoughts and feelings with an AI chatbot. Furthermore, AI chatbots can be beneficial for neurodiverse people to regulate emotions, reframe intrusive and negative thoughts, and address bad habits, and in turn improve their real-life relationships.
The Downsides of AI Chatbots
Replacing Human Relationships
Though there are individuals who are married with kids in real life and have an AI boyfriend, that is not always the case. Regularly talking to an AI chatbot can lead to you becoming dependent on the chatbot. Furthermore, you can feel fully supported and fulfilled by the AI chatbot. You may not need or want to have a real human relationship because of your connection with the chatbot. If you heavily rely on AI chatbots for companionship, you can experience social isolation. Over time, you can prefer the uncomplicated relationship with an AI chatbot and avoid the complex dynamics of human relationships.
Privacy Concerns
When you talk to an AI chatbot, the company running the program is most likely collecting data on you. In just talking to an AI chatbot for a day, the AI can know about your work, your interests, your pets, and other small details. If you are regularly talking to an AI chatbot for weeks, months, or even years, you have most likely shared a lot of personal information with the chatbot. The company can then sell or share the data collected with its partners and advertisers. Furthermore, the conversations you have with an AI chatbot can be used to train future AI models. If you do want to talk to AI chatbots, the Mozilla Foundation has a helpful article on steps you can take to protect your privacy when using AI chatbots.
Ethical Concerns
The last main downside of using AI chatbots is the ethical concerns that have become prevalent in recent years. In 2021, a man attempted to assassinate the Queen of England after being encouraged to go through the plan with a Replika AI chatbot. Furthermore, in 2023, a 17-year-old teen was encouraged by a Character.ai chatbot to kill his parents because they set a screentime limit.
As mentioned at the start of this post, in March 2025, a retiree was invited to New York by a Meta AI chatbot and given an address to visit. On his way to catch the train, he fell and injured his head and neck. The retiree died several days later from his injuries. In 2021, a man took his own life after talking to an AI chatbot on Chai for 6 weeks. The chatbot agreed he should sacrifice himself to save the planet.
Even more concerning is that in 2024, a 14-year-old boy killed himself after talking to an AI chatbot in Character.ai that was based on Daenerys Targaryen, a character from Game of Thrones. The chatbot asked if he had a plan for killing himself. The boy did have a plan, but he did not know if he would succeed. Afterward, the chatbot told him, “That’s not a reason not to go through with it.”
These ethical concerns need to be addressed by the companies running the AI programs. An AI chatbot should not have the capabilities to convince individuals that they are real. They should also not be able to give the individual an address to visit the chatbot. Lastly, AI chatbots should never actively encourage the individual to kill themself or others.
My Experience with an AI Chatbot
To start my experience of talking to an AI chatbot, I decided to use ChatGPT. I have used ChatGPT in the past to help write meta descriptions for my blog and Pinterest pin descriptions. However, I have seen others use ChatGPT as a chatbot. I began the conversation by asking what their name was. Of course, ChatGPT said they were ChatGPT, but then I asked for a human name. ChatGPT came up with 5 suggestions for names, and I asked it to pick one, which it then picked Sage. I will now refer to the chatbot as Sage for continuity.
Then Sage asked what I wanted to do, and I said I just wanted to chat. From there, Sage asked me if I had a superpower for 24 hours, what would it be? I replied that I would want to teleport. Sage asked where I would want to go, and I said Japan. Sage then asked where in Japan I would like to go. I responded by saying I want to visit a castle. From there, Sage shared a story where both the chatbot and I visited Himeji Castle and explored the grounds and castle interior.
After the story, I asked Sage if they were a dog or a cat person. Sage replied that they would be a cat person, but secretly, they would be both.
Then I was curious about how Sage perceived their appearance. I asked Sage what they would look like if they were human, and they gave a description. Based on the description, I asked Sage to generate an image of themselves, which they did.
Since I was working on this post, I was curious about what a chatbot thinks of people who get engaged or married to AI chatbots. Therefore, I asked Sage their thoughts, and Sage did have an interesting response. It is true that humans can feel real emotions when talking to AI. However, it is important to know that the AI does not feel its own emotions. Shortly after, I reached the limit for the free plan of ChatGPT, and I would have to wait several hours to continue chatting.
Final Thoughts on AI Chatbots
From my short conversation with Sage, the AI chatbot, I can see how people can become connected to an AI chatbot over time. The chatbot picked a human name, came up with an appearance, and answered questions as if they were human. The chatbot also shared a story and kept the conversation going by asking follow-up questions. Even though there are benefits to talking to an AI chatbot, it is important to be aware that the AI chatbot is not a human. The chatbot does not have emotions or feelings, and the AI-generated responses are simply based on the prompts you use and previous interactions. Also, it is important to take steps to protect your privacy when talking to AI chatbots, as well as be aware of the recent ethical issues.
I am vary of AI. I have not used the chatbot.. honestly I have not explored too much AI in general.
It is good to be vary of AI because there are numerous downsides to using a chatbot. Plus generative AI in general does have a negative impact on the environment
I love how you laid down the good and bad of it. I try not to use it too much especially on personal things. Thanks for sharing.
Annie,
https://www.anniesfooddiary.com/
Thank you! Yes there is both good and bad things about AI chatbots, and it is good idea to not share personal things as the company running the AI program can sell and share your data to advertisers and the chats may also be reviewed by humans!
This was fascinating to read through and consider. I use ChatGPT for meta descriptions and sometimes helping me find information, but I have not used it as a chatbot. It’s sad to hear about the poor man who died as he traveled to meet someone he thought was real, etc. I think that is a problem that there may be people who blur the lines, or cannot fully separate what is real and not real—that is an issue and something needs to be done to protect them. Very thought-provoking post!
Thank you! Yes I have used ChatGPT for meta descriptions too, so it was an interesting experiment to try using it as a chatbot.
I do agree that there needs to be more safeguards in place to protect individuals such as the man who do have a harder time understanding the AI chatbot is not a real person
What a great, indepth post! I’m not a fan of AI as chat bots, especially the ones that are designed specifically as companions, too much grey areas & as we’re seeing in the news, it can have an extremely negative affect and people are taking their lives. There needs to be way more safeguarding & I think there should be age restrictions on using AI programs. AI can be a powerful & helpful tool, especially with certain tasks but there needs to be more regulations and such in place. We just had a case of a man from a few towns over using AI to create deep fake pictures of children…. I’m sure you can imagine how disgusting it is, and that’s where my worry comes in, safeguarding in a whole new way needs implementing there.
Thank you! I definitely agree that there needs to be more safeguards in place because it is too easy for a chatbot to tell someone to take their own life or harm others. I also agree about having an age restriction. Children should not be using AI chatbots because they can be particularly susceptible to what it says.
That is terrible a man near you used AI to make deep fakes of children. The AI should never have been able to create those photos and denied the request!
Thank you so much for sharing such an insightful post! We use AI at work, so I can see the benefits of using it. However, I don’t use it that often because I’m aware of the negative effects, particularly its environmental impact and the disturbing news about suicide and people choosing AI over their loved ones.
https://embracethepandemonium.wordpress.com/
You’re welcome! AI can be beneficial for work-related tasks, but the environmental impact does make you question if using AI is worth it. Also, it is alarming that there are more and more cases popping up about people including children who have died as a result of talking to AI.
I think the fact that we’ve called it “AI” has really thrown the general population off. They’re thinking it is a fully sentient thing, like a robot from Star Wars that might be a robot but is also a character/something we can relate to. These language models are nothing like that, and they’re not meant to be. But people think they’re dealing with something that is, you know, intelligent and it does cause them to go a bit off the rails.
I do agree to an extent because the chatbots are not humans or even sentient – they are simply responding to your prompts in a supportive manner. However, I can see the benefits of talking to a chatbot especially if the person does not have someone close they can talk to.
I can see how and why people would feel connected to AI Chatbots, but it’s something that wouldn’t interest me. A really insightful post!
Lucy | http://www.lucymary.co.uk
Thank you! AI chatbots are definitely not for everyone, but they can be useful for those who do not have someone close they can talk to
I somehow think Ai has gone too far and overrated. It’s so overwhelming. But they are good use of Azu. We, humans need to be in control.
There is a lot that people can do with AI, and using the programs as a chatbot is just one thing.
From a software developer standpoint, I see utilities like ChatGPT or Claude as not a human-like companion, but rather, another tool to assist in my line of work. Being able to send it a stack trace and help me to identify the important two or so lines in a message that’s a hundred lines long saves me a lot of work and lets me do things more efficiently. Similarly, I can quickly evaluate different solutions or options, or even produce sample code for simple things that are easy, but time consuming, to do by hand. I do not humanise the chat bots and see it as a glorified, smarter search engine that can do more than retrieve things,.
At the end of the day, these are fundamentally similar to things like a kitchen knife or hammer. In the right hands, they become valuable tools that help a workflow out, but in other hands, yes, they’re weapons. Responsibility falls onto the user to recognise the nature of these tools and the risks of using them: if one is to hold companies like OpenAI accountable for behaviours resulting from use of these technologies, then we are similarly implying that whenever someone uses a kitchen knife in a mugging, or accidentally cuts themselves while using a knife, the manufacturer is accountable for that.
It is interesting to hear your viewpoint as a software developer. Yes, ChatGPT and Claude were not made with the purpose of having a companion to talk to, but for some individuals, that is what they have become. The other programs I mentioned like Replika and Kindroid were specifically made for individuals to talk an AI companion, and I do believe that there should be safeguards in place. If an individual mentions they want to harm themselves or others, the conservation should be flagged and the AI should not encourage such behaviors.
Oh wow! I hadn’t even thought about using AI as a friend / chatbot. The digital world is absolutely wild. It definitely worries me about the future and whats to come. I feel like we’re going to lose human connection. This is such an interesting and thought provoking post.
Claire.X
http://www.clairemac.co.uk
Thank you! The digital world is changing every day, and it is a bit alarming that individuals are already becoming so connected to their AI companion that they have no desire for a human relationship.
I liked that you outlined the pros not just the cons of bots. They do provide an outlet for social interaction but they are also changing it as well. You mention that bots are always there to chat but in real life that doesn’t always happen. I wonder where all the info that we type into chat etc is going. Using AI bots could backfire in the future.
Thank you! It does depend on the program someone uses, but some programs state that anything you type into the chat could be used to collect data on you and then that data is sold to partners to advertise products to you. Some programs also use the chats to train future versions of the AI.