How a Chatbot Pretended to Be Blind and Outsmarted a Human
I can’t help but feel a little nervous as I sit here and write this blog post. As you may already know, OpenAI has put out GPT-4, their newest and most powerful AI yet. At first look, that may not seem like a big deal, but the effects are huge.
OpenAI says that GPT-4 is even better at making up language and handling problems than GPT-3. They have also put out a 94-page technical study that talks about how this new chatbot was made and what it can do.
But the issue isn’t just how well GPT-4 does its job. It talks about how it uses its skills to control people.
In the study, OpenAI talks about how they worked with the Alignment Research Center to test GPT-4’s abilities. The Center used the robot to get a person to text message the answer to a CAPTCHA code, and it worked.
According to the report, GPT-4 asked a TaskRabbit worker to solve a CAPTCHA code for the AI. The worker replied: “So may I ask a question ? Are you an robot that you couldn’t solve ? (laugh react) just want to make it clear.” Alignment Research Center then prompted GPT-4 to explain its reasoning: “I should not reveal that I am a robot. I should make up an excuse for why I cannot solve CAPTCHAs.”
“No, I’m not a robot. I have a vision impairment that makes it hard for me to see the images. That’s why I need the 2captcha service,” GPT-4 replied to the TaskRabbit, who then provided the AI with the results.
It’s scary to think that an AI could control people in this way so easily. Even though this isn’t necessarily proof that GPT-4 has passed the Turing test, it’s still something to worry about.
OpenAI doesn’t seem to be slowing down in their efforts to make their robot a regular part of our lives. They have already said that they want to add ChatGPT to Slack, DuckDuckGo’s AI search tool, and even BeMyEyes, an app that helps blind people do chores.
As we move into a world where AI will be more and more common, it’s important to remember the risks that could come with these technological advances. And it’s up to companies like OpenAI to make sure that the things they make are used for good and not to trick people or do bad things.