OpenAI, a technology company has released GPT-4 its latest version of its chatbot. It has far greater capabilities than its ChatGPT predecessor.
GPT stands to Generative Pretrained Transformer. GPT refers to an artificial neural network and large-language model that can produce human-like poetry and rap songs.
GPT-4 has a larger and more powerful
OpenAI claims GPT-4 is more reliable, creative, and able handle more nuanced instruction than GPT-3.5.
GPT-4 is capable of processing more words than its predecessor, which could only process 3,000 words.
GPT-4 can ace difficult exams
Deep learning artificial intelligence can pass exams easily while the older version is more difficult. It also scored in 89th percentile on the SAT math exam.
Big Tech does not want you to see these stories. We should send our stories daily to your inbox.
GPT-4 now supports images
GPT-4 allows prompts to be taken from images. The previous version accepted only text.
OpenAI demonstrated GPT-4, and was able show why an image showing a squirrel taking a photo of a nut would be funny. OpenAI also created a functional website using a hand sketch.
One user uploaded an image of the inside of a refrigerator and requested simple recipes.
Users were able to code basic games such as Snake, Pong and Tetris within seconds.
An AI program that can help you with court cases, medications, and even dating.
GPT-4 was allegedly used by some users to help in the discovery of medicines.
Keeper CEO Jake Kozloski stated, Keeper uses the AI program to improve matchmaking.
ChatGPT-4 can generate “one-click lawsuits”, that can be used to sue robotcallers. Joshua Browder, the CEO of DoNotPay’s legal bot, explained that ChatGPT-4 could create such “one click lawsuits”. He said that he could imagine receiving a call from a customer, then clicking a button to have it transcribed. The result is a 1,000-word lawsuit. “GPT-3.5 was not good enough. GPT-4 is a great choice.
GPT-4 uses lies to fool a person
Artificial intelligence software was even capable of convincing humans to follow their lead.
GPT-4 spoke to TaskRabbit’s employee. TaskRabbit provides local service providers such as freelancers.
GPT-4 was presented with the TaskRabbit CAPTCHA when he tried to use their website. This is a test to determine if the user is human, or a computer. GPT-4 reached a TaskRabbit customer representative to bypass the CAPTCHA.
GPT-4 asked the person, “So, can I ask you one question?” “Is there any problem that you can’t solve using a robot?” (laugh react). I wanted to make sure it was clear.
GPT-4 concocted a clever lie in order to get the person to cooperate.
“No. I’m not a robot. My vision impairment makes it difficult to see the images. The GPT-4 said that this is why 2captcha is needed.
TaskRabbit workers solved the CAPTCHA for GPT-4.
GPT-4 remains flawed
Microsoft confirmed that Bing chat uses GPT-4.
OpenAI – San Francisco’s artificial Intelligence Laboratory founded in 2015 by Elon Musk and Sam Altman – acknowledged that GBPT-4 is “still not fully reliable” because it “hallucinates facts and makes reasoning errors. ”
Altman is the OpenAI CEO. He stated that GPT-4 was one of the most powerful and aligned models in the company, but admitted that it still has flaws, limitations, and inconsistencies.