Generative AI - Short & Sweet 12 - GPT-4 ??
"Multimodal AI models are opening up new dimensions of creativity, enabling machines to see, hear, and speak in ways never before possible." - ChatGPT (GPT-3.5)
???The Generative AI topics of the week are ...
GPT-4.?ChatGPT which has GPT-3.5 in its core had a worldwide ripple effect. Its successor GPT-4, a large, multi-modal model outperforms given models on pretty much all verticals, and has other capabilities that leave us in awe.
Creativity
GPT-4 is more creative and collaborative than ever before. It can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style.
Multi-modal
GPT-4 accepts images as input and generates captions, classifications, and analyses in painstaking detail.
Extended outputs
GPT-4 is capable of handling over 25,000 words of text ( roughly a quarter of a 270p. book), allowing for use cases like long-form content creation, extended conversations, and document search and analysis.
Long context input
GPT-4 is capable of accepting long contextual input information. It accepts up to 32k tokens, which is roughly 43000 words, which is roughly half of a 270p. book.
In the demo yesterday March 14th, 2023, they demoed how a napkin sketch is transformed into a functional webpage.
In another example, they were telling the system to behave as a tax adviser. It should carefully read & apply the tax code, being certain to spell out all calculations & reasoning so anyone can verify them.?
The result is stunning.?
OpenAI has tested GPT-4 on a variety of benchmarks, including simulating exams that were initially designed for humans. They proceeded by using the most recent publicly-available tests (in the case of the Olympiads and AP free response questions) or by purchasing 2022–2023 editions of practice exams. They did no specific training for these exams.
领英推荐
These are the results. GPT-4 could be accepted at any Ivy League university.
And sure there is also a dangerous side to it. See Twitter. In the meantime, Skynet-like scenarios are being envisioned.?
However, at this point in time, GPT-4 and similar models are tools, like knives, that can be used in unethical and harmful ways, but also bear incredible upside potential for all of us.?
How to get access?
Watch yesterday’s developer demo here [2].?
Subscribing to, giving feedback about, and sharing the?newsletter?as well as our?renowned online course?will be highly appreciated and helps a lot. ??
Do you know about GenAI Gems ?? that we should consider? Or other matters??
Please, respond to this email.
Thank you for reading,?
Martin?
References:
STSM, watsonx Code Assistant
2 年Thanks for the concise summary! Suggestion : ditch the stereotypical female robot. I'm getting tired of seeing body shapes from times where women were tied and bound into corsets...
Empowering AI-First Enterprises with LLMs & Agentic AI: A 35-Year Veteran for Enterprise AI.
2 年Excellent summary, Martin
Head of AIOps @ IBM || Speaker | Lecturer | Advisor
2 年Very insightful. Thanks for sharing.