OpenAI Announces GPT-4: An Upgrade to ChatGPT with Improved Accuracy and Problem-Solving Skills

OpenAI has just announced the highly-anticipated GPT-4 language model, an upgrade to the foundational technology behind the ChatGPT system that promises to bring to light the firm's “most advanced system, producing safer and more useful responses.”[0] GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.[1]

Microsoft confirmed that GPT-4 is available to ChatGPT Plus subscribers, otherwise users will have to sign up for the waitlist to test it out.[2] Andreas Braun, Chief Technology Officer at Microsoft Germany, revealed that GPT-4 will include multimodal models, such as videos.[3]

GPT-4 has improved accuracy, problem-solving abilities, and reasoning skills according to OpenAI.[4] It scored in the 90th percentile on the bar exam versus the 10th percentile with GPT-3, and 99th in the Biology Olympiad versus GPT-3 which scored in the 31st percentile.[5]

However, the model is still flawed.[6] It will continue to imagine, creating invented information.[6] OpenAI has observed that hallucination could be a greater concern as GPT-4 does not do this as often, thus people might be complacent with the outputs it produces.[6] The model can still be made to produce biased and harmful language.[6] OpenAI has divulged little information about the size of GPT-4, the number of GPUs used to train it, or what data was used for training.[6]

ChatGPT has become wildly popular, becoming the fastest-growing consumer app in history to reach 100 million users.[7] Microsoft and Google aim to embed content-creating generative AI technologies into their internet search and office productivity products.[8] OpenAI is saying that GPT-4 is still limited in terms of its ability to understand and respond to nuanced instructions, and while it has improved accuracy, problem-solving ability and reasoning skills, it still has many known limitations such as social biases, hallucinations, adversarial prompts and lack of knowledge of events that happened after September 2021, when its training data was finalized, and “does not learn from its experience.[9]

0. “OpenAI says new model GPT-4 is more creative and less likely to invent facts” The Guardian, 14 Mar. 2023,

1. “OpenAI releases newest artificial intelligence tech: GPT-4” AOL, 14 Mar. 2023,

2. “OpenAI releases GPT-4, a multimodal AI that it claims is state-of-the-art” Yahoo Life, 14 Mar. 2023,

3. “Microsoft will launch ChatGPT 4 with AI videos next week” Digital Trends, 10 Mar. 2023,

4. “OpenAI Releases GPT-4: Now Available In ChatGPT & Bing” Search Engine Journal, 14 Mar. 2023,

5. “OpenAI announces GPT-4” Mashable, 14 Mar. 2023,

6. “GPT-4 debuts and Google beats Microsoft in race to add generative A.I. to consumer office tools” Fortune, 14 Mar. 2023,

7. “OpenAI Announces Chat GPT-4, an AI That Can Understand Photos” PetaPixel, 14 Mar. 2023,

8. “OpenAI Launches Multimodal GPT-4, Adding To Growing Buzz Over AI Stocks” Investor's Business Daily, 14 Mar. 2023,

9. “OpenAI releases major new chatbot engine” Axios, 14 Mar. 2023,

Click Here to Leave a Comment Below 0 comments