The debate around conversational and generative AI is heating up as the premier search engines of the world, Google and Bing, incorporate AI assistants into their search functionalities. However, as useful as these AI systems can be when looking for information online, many users have reported increasingly weird responses.Online communities relating to ChatGPT and Microsoft's Bing AI are jam-packed with threads containing screenshots of the Bing AI delivering some outlandish and downright horrendous responses. Other AI systems, including OpenAI's wildly popular ChatGPT, haven't been spared.But all that you see isn't true, and you shouldn't trust every AI chatbot screenshot you see.

Are AI Chatbots Broken?

The first thing these screenshots might make you believe is that AI chatbots are broken. No one's going to think highly of generative AI bots like ChatGPT after seeing them struggle to respond to user queries and going on rants about how they don't really know what they're doing.

As impressive as ChatGPT, Microsoft's Bing AI, and Google Bard are, they aren't perfect and will break, especially when millions of people worldwide are trying out all sorts of different queries.

There are several ways AI chatbots can be broken. Even asking something as simple as rephrasing their current answer can confuse the bot. You can up the ante by asking it anything existential or emotional, and you'll be in for a ride, and we've covered some of these methods to get the best out of ChatGPT.

To put it all in perspective, these systems are continuously being trained, monitored, and tuned to churn out better and better responses to the point where they might become a threat to search engines themselves. A query that might break them today might be answerable tomorrow. As the companies behind these systems work to push the boundaries of what AI can do, these chatbots will keep getting better. They're not broken; it's in their nature to evolve continually.

Why Can't Screenshots Be Trusted?

Screenshots, in their most basic form, are simply images of a conversation, and images can be tampered with rather easily. Changing a screenshot to present a different narrative isn't something new either. All it takes is five minutes and a few basic Photoshop skills to turn a conversation on its head and turn a regular screenshot into a sensational Reddit post.

The AI chatbot trend, or any trend for that matter, can drive internet traffic. Getting your hands on an interesting screenshot that can be paired with a clickbait title could deliver an insane amount of clicks, which might give someone the incentive to get creative with their screenshots.

To demonstrate, here's a quick example of a real and fake conversation between ChatGPT and me.

What to Keep in Mind When Seeing AI Screenshots on the Internet?

First, take everything with a grain of salt, especially if the source isn't reputable. Any Reddit, Instagram, Twitter, or social media posts featuring AI screenshots aren't exactly the most reputable source for you to trust what you see. This also extends to any lesser-known websites or blogs that might be trying to get on the AI bandwagon to get more traffic.

Other than the source, try to spot visual artifacts in the image itself. Badly edited screenshots often tend to have mismatching fonts or misplaced UI elements that can give away the authenticity of an image. There's no guaranteed way of telling whether an image has been tampered with, but visual artifacts often are the biggest giveaway here.

Finally, ChatGPT is publicly available, and Bing is working hard to roll out its new search assistant to as many people as possible. Google also revealed its Bard AI assistant, but its first demonstration wasn't clean sailing. If you see a prompt with an unreasonable or shocking answer, run the query yourself. The answer might not be the same as you saw in the screenshot, but it will give you a better understanding of whether or not the screenshot was true.

AI Chatbots Don't Really Want to Take Over the World

AI chatbots are rather helpful, but they aren't perfect. Occasionally, a prompt will generate some weird responses, but that isn't reason enough to write them off.

The advantage here is just too big to ignore. If you can fine-tune your prompts, you can harness these AI systems' power and make your life easier.

Just remember, not everything you read on the internet is true.