Introducing ChatGPT Paid Features (October 2023)

English summary

This video, dated October 14, 2023, introduces and demonstrates major ChatGPT Plus features. The features covered are GPT-4, GPT-4V image input, Bing web browsing, Advanced Data Analytics, plugins, DALL-E 3, and voice conversations.

Key points

• GPT-4 is presented as providing higher response quality, reasoning, contextual understanding, and compliance with Korean-language constraints than GPT-3.5. The presentation states that GPT-4 supports a context window of about 4,000 tokens and recommends drafting large documents in English and translating them afterward because Korean uses more tokens than English.
• GPT-4V is demonstrated for receipt calculations, multilingual and handwritten text recognition, educational diagram explanation, mathematical problem solving, medical image interpretation, and explanation of foods and meme context.
• Bing web browsing is introduced as a way to retrieve information after the September 2021 knowledge cutoff and to provide clickable source links; a 2023 revenue comparison involving OpenAI, Kakao, and Naver is demonstrated. Advanced Data Analytics is described as running Python in a sandbox on uploaded CSV or Excel files for exploratory analysis, regression and random forest modeling, RMSE comparison, prediction plots, feature importance analysis, and downloadable outputs. The demonstration uses a clinical warfarin dosage dataset with variables including CYP2C9 and VKORC1. Automatic code correction and re-execution after runtime errors are also demonstrated. Plugins are presented for diagram generation, automation, mathematical computation, and document querying. DALL-E 3 is shown generating illustrations from internally expanded prompts, while voice conversations are presented for language practice and interactive storytelling.

Audience

Academic researchers, thesis writers, university students, data analysts, educators, and AI practitioners evaluating whether to upgrade to ChatGPT Plus.

Scope and limitations

This summary reflects only the feature descriptions and demonstrations reported in the October 14, 2023 video notes. The demonstrations do not constitute independent validation of general performance or medical-image diagnostic accuracy; the video explicitly notes that the vision model can misdiagnose or fail on complex medical cases. The demonstrated interface also required users to select one tool mode at a time, GPT-4 and web browsing could be slower than GPT-3.5 and standalone Bing Chat, and plugin availability and feature configurations may have changed after the reference date.

Research context

Within a thematic connection to Sangzin Ahn's portfolio, this material relates to medical AI and data science through examples of generative AI, multimodal input, web-based information retrieval, clinical dataset analysis, model comparison, and research artifact generation.