AI is advancing at unprecedented levels, with new developments happening almost every week. Generative AI tools like ChatGPT are now very popular and integrated everywhere.
But should it? Using AI technology for productivity, education, and entertainment makes sense. But companies are now looking to embed it directly into their messaging apps, which could have disruptive consequences. Here are 7 reasons why.
1. AI chatbots tend to hallucinate
If you’ve used ChatGPT, Bing, or Bard, you know that generative AI chatbots tend to “hallucinate”. AI hallucinations are invented by chatbots due to a lack of proper training data on user-requested queries.
In other words, they misinform, but sound confident as if it were the case. This is a big problem because many people don’t do fact-checking when using chatbots and believe chatbots are accurate by default. This is one of the biggest mistakes to avoid when using AI tools.
When incorporated into messaging apps, it allows people (intentionally or unintentionally) to spread misinformation between contacts and on social media, spread propaganda, and facilitate echo chambers. Because of the possibility of using it, the amount of harm it can do is even greater.
2. People don’t like talking to bots
Considering the annoyance of trying to contact a company’s customer support and having to speak to a chatbot instead of a live human executive who can actually understand the nuances of the issue and provide the right advice. please look.
The same applies to personal conversations. Imagine you are talking to a friend. Along the way, they realize they’ve been using AI to respond to your messages all along, instead of responding themselves based on their own thoughts and opinions.
If you’re like most people, the use of AI in private conversations can quickly become offensive, callous, and creepy, as if the other person doesn’t think you’re worthy of their time or attention. and even perceived as passive-aggressive. sympathy.
For example, using AI to compose emails is understandable as it is a professional interaction, but using AI in personal conversations is something no one would want to recommend. As the novelty of the technology wears off, it would be rude to use it in this context.
3. AI cannot copy your unique tonality
Today’s generative AI tools already allow you to change the tone of your message, whether it’s formal, playful or neutral, depending on who you want to address and how you want to say it. For example, you can use Magic Compose in Google Messages to do the same.
That’s good, but keep in mind that these dialects are trained based on the training data you set, not your personal chat history, so you can’t duplicate your unique dialect or commonly used emojis.
Especially if you’re using AI to compose simple work emails where everyone uses more or less the same formal tone, you might not care much about this. But when you use it to talk to friends and family in the Messages app, it matters a lot more than you might think.
Until you allow the AI tool the option to train a language model based on your chat history, the AI tool will not be able to reproduce your unique dialects and quirks. That said, solving this challenge isn’t that hard, so it might be implemented soon.
4. Creating good prompts takes time
Achieving desirable results from AI chatbots is highly dependent on the quality of the prompts. Improperly crafted prompts will return bad responses, so you should modify your prompts until you are satisfied with the results.
This process makes sense when you want to write long-form content, but it’s very inefficient when writing multiple short responses in informal conversation.
In most cases, the time it takes to refine a prompt and get a usable response is longer than it would have taken to compose the message yourself.
5. AI can produce aggressive results
Apart from accuracy, bias is one of the biggest problems in generative AI. Some people perceive AI to be unbiased because it has no motivation of its own. But these AI tools are ultimately powered by humans with their own biases.
In other words, the bias is built into the system. AIs inherently do not understand what is considered offensive and what is not, so they are trained to be prejudiced against, for example, certain groups of people or certain cultures, As a result, it can produce aggressive results in the process.
6. AI may not understand sarcasm and humor
AI’s understanding of metaphorical expressions, such as sarcasm and metaphors, has improved over time, but it is still far from being able to recognize humor in conversation. For example, when I sarcastically asked his Bard at Google, the results were hit and miss.
In some cases it was really funny and played to my sarcasm. Other times, however, I’ve reverted to defaulting to uninteresting cookie-cutter responses, or simply refusing to participate in the conversation, saying that being just an LLM doesn’t help my question.
7. Reliance on AI can lead to poor communication
Another subtle but crucial question when integrating generative AI into messaging apps is how it will affect our ability to communicate. As we increasingly rely on AI to talk to each other, it could hamper our ability to train our emotional intelligence and social skills.
The point here is that the more societal needs are outsourced to AI, the harder it will be to communicate ideas in an organic way. In other words, the more you use AI to talk to your contacts, the more likely it is that the quality of your relationship will suffer.
Not everything needs AI
With the advent of new technology, we are often too busy thinking about how to use it to discuss whether we should use it in the first place.
While it makes perfect sense to use generative AI to compose emails, brainstorm ideas, or create images for presentations, its integration into messaging apps invites a lot of criticism.
