I must admit, until a few months ago, I hadn’t even heard of GPT, and it’s only very recently that I bothered to find out what the acronym stands for.
GPT, or Generative Pre-trained Transformer, is a series of advanced artificial intelligence models developed by OpenAI.
The GPT models are a type of deep learning language model, specifically using the Transformer architecture, which was introduced by Vaswani et al. in 2017. GPT models are designed to generate human-like text based on a given input, and they can be fine-tuned for various natural language processing tasks, such as text generation, translation, summarisation, and question-answering.
Via https://chat.openai.com/chat?model=gpt-4
I started exploring OpenAI’s Chat, initially on GPT-3, and was very impressed. When I heard that a new version was coming out, I watched the announcement, and immediately subscribed to Chat GPT Plus, in order to get hands-on with the latest version, GPT-4.
Here’s one of the first questions that I asked GPT-4:
Please suggest some metaphors for each of the different versions of GPT(including GPT-4).
And here’s the answer I was given:
- GPT-1 as a budding wordsmith: The first version of GPT is like an aspiring wordsmith, showcasing initial talent in natural language understanding and generation, but still limited in its abilities and requiring further refinement.
- GPT-2 as an apprentice storyteller: GPT-2 is like an apprentice storyteller, demonstrating a significant improvement over its predecessor. It’s capable of generating more coherent and contextually appropriate text but still sometimes fumbles and requires guidance.
- GPT-3 as a skilled novelist: GPT-3 is akin to a skilled novelist, with an impressive command of language and the ability to generate creative, contextually rich, and relevant text. However, like any author, it occasionally makes mistakes or gets lost in the details.
- GPT-4 as a hypothetical master orator: If GPT-4 were to be developed, it might be like a master orator, building upon the strengths of its predecessors to generate even more sophisticated, accurate, and engaging language. As a hypothetical version, GPT-4 could be expected to have an even deeper understanding of context and nuance, while minimising errors and offering new capabilities.
Please note that GPT-4 is a hypothetical version of the GPT series and has not been released as of my knowledge cutoff date in September 2021. The metaphor for GPT-4 is a projection of what it might be like based on the improvements seen in previous versions.
I highlighted the caveat above. OpenAI’s Chat GPT-4 is extremely impressive, but it’s worth noting that it has no awareness of anything since September 2021.
That’s not the case with the new Bing, from Microsoft. Although Bing is based on GPT-4, it can also search the web, and can therefore include current information in its responses on Chat.
Can you suggest another metaphor for each of the different versions of GPT(including GPT-4). Can you try to include a metaphor for Bing AI Chat alongside please?
Here’s what Bing AI Chat came up with:
Sure, here are some metaphors for each of the different versions of GPT:
- GPT-1: A baby that is just learning to talk.
- GPT-2: A child that is starting to read and write.
- GPT-3: A teenager that is very smart and can help with homework.
- GPT-4: A college student that is an expert in many subjects and can help with research papers.
As for Bing AI Chat, it is like a personal assistant that can help you with anything you need. It is like having a friend that knows everything and is always there to help you out.
That’s all for this post. I am writing these posts primarily as a memory-jogger for myself, but thought that posting them publicly might interest others as well.
If you find any of this useful – or you think I’m getting it completely wrong, or have tips or suggestions – please let me know in the comments below.
Discover more from Mark Braggins
Subscribe to get the latest posts sent to your email.