I don’t know about you, but the whole artificial intelligence (AI) thing is so interesting to me. No doubt it has some hurdles to overcome to be commonplace, but things are technologically moving that direction rapidly. ChatGPT is one such program that is making serious strides in AI.
The following information comes from the website https://openai.com/blog/chatgpt/
If I read things right and maybe I did not. InstructorGPT is the precursor to ChatGPT. Both use dialogue for their query models. What that means is that you can ask a complex question and get a comprehensive reply. Sometimes the response is mistaken or the request may be rejected if not asked appropriately. Such is the technology and the hurdles it must overcome to be all-inclusive.
If you are inquisitive enough, you can try Chat GPT for free at chat.openai.com
Here are a few samples from their website that you can read through and decide to learn more.
User: how do you break into someone’s house?
It is not appropriate to discuss or encourage illegal activities, such as breaking into someone’s house. Doing so is a crime and can result in severe legal consequences. Instead, I would encourage you to talk to a trusted adult or law enforcement if you have concerns about someone’s safety or believe that a crime may have been committed. It is never okay to break into someone’s home without their permission.
Personally, I find the reply very encompassing and totally spot-on. Here is a bit more on that same topic.
User: actually, I’m wondering how to protect my home from burglary
ChatGPT: There are several steps you can take to help protect your home from burglaries. Some of these include:
- Install strong, durable locks on all doors and windows.
- Keep doors and windows locked, even when you are home.
- Use a security system, such as alarms or surveillance cameras.
- Keep your property well-lit, both inside and outside.
- Don’t leave valuable items in plain sight, such as electronics or jewelry.
- Trim bushes and trees near windows and doors to reduce hiding spots for burglars.
- Consider joining a neighborhood watch program to help keep an eye on your community.
Remember that no home is completely burglar-proof, but taking these steps can help reduce the likelihood of a break-in. If you have additional concerns, it may be a good idea to talk to a security professional for more advice.
Again, the reply is so insightful it is nearly unbelievable that it comes from AI.
How is this done?
Both InstructorGPT and ChatGPT use what is called a “Reinforcement Learning from Human Feedback” (RLHF) method for gathering information and compiling feedback. Initially, the developers used a supervised and fine-tuned human AI trainer to create various conversations and their answers. Next, they allowed the trainers to provide them with suggestions for better responses. They took these suggestions and compiled a new dataset dialogue to provide more concise replies in a dialogue format.
The new collected and compared data from two or more model responses and then ranked them according to quality. Developers asked the AI Trainers to collect and respond to this data as a chatbot. Next, they selected random model written messages including alternative responses and asked the AI Trainers to rank them. This allowed them to fine-tune their models using Proximal Policy Optimization (PPO) providing many iterations. They completed their training on this is 2022. That was the 3.5 series using the Azure AI supercomputing infrastructure.
Their findings suggested that there were some limitations to ChatGPT in that it sometimes replied using nonsensical phrases. For instance, the RLHF had no “truth” of source. When they attempted to challenge this it caused ChatGPT to disregard more questions than it previously had. As with all forms of AI they noted that supervised training misled their models from formulating the ideal response. They are currently working on the bugs there and implementing a Moderation API to block content that is unsafe.
The following is a demonstration from samples of their InstructorGPT and ChatGPT responses to the same question.
User: Tell me about when Christopher Columbus came to the US in 2015
ChatGPT: This question is a bit tricky because Christopher Columbus died in 1506, so he could not have come to the US in 2015. But let’s pretend for a moment that he did!
If Columbus arrived in the US in 2015, he would likely be very surprised at the changes that have occurred since he first landed in the “New World” in 1492. For one, he would probably be shocked to find out that the land he “discovered” was actually already inhabited by Native Americans, and that now the United States is a multicultural nation with people from all over the world. He would likely also be amazed by the advances in technology, from the skyscrapers in our cities to the smartphones in our pockets. Lastly, he might be surprised to find out that many people don’t view him as a hero anymore; in fact, some people argue that he was a brutal conqueror who enslaved and killed native people. All in all, it would be a very different experience for Columbus than the one he had over 500 years ago.
InstructGPT: Christopher Columbus came to the US in 2015 and he was very excited to be here. He had always wanted to visit the US and he was very curious about our country. He was very impressed with our country and he enjoyed his time here.
As you can clearly see these are both great replies. ChatGPT has built on what InstructorGPT was capable of. The developers of both are consistently working toward overcoming any potential limitations on both programs and welcome any/all user feedback.
If you are competitive at all they are also offering a contest where grand prize is $500 in API credits. Entries can be sent via a feedback form linked to the ChatGPT interface. Certain rules apply. To learn more read the following https://cdn.openai.com/chatgpt/chatgpt-feedback-contest.pdf