Thank you! I think this comes from my most recent position working in an elementary school STEM lab. I had to constantly adapt lessons to different age groups and learning styles, (the kids in the "gifted and talented" program needed a different sort of learning approach than the kids in the standard afterschool program.) I took it as a personal challenge to engage all of these groups, all of the time.
When I write AI articles, I am acutely aware I have a very diverse audience. Some of my readers have a very high-level knowledge of AI-terminology. But, some readers are new to AI, but very much want to learn more. I also have readers who speak English As a Second Language. To meet the needs of all these groups, I added the Vocab Key.
In the event my writing was not clear, I added the FAQs. The FAQs reinforce the main points of the article, but worded in a different manner.
In the event someone prefers to listen rather than read, I offer the podcast.
In the event someone is a visual learner, I offer the charts and diagrams.
It takes a lot of energy to do the articles. I always tell my husband each one is like "giving birth." But, I always feel a great sense of accomplishment when I am done.
This is a great question, and it depends upon which model of DeepSeek you are referring to. The easiest way to answer the question is to simply ask DeepSeek. I access DeepSeek through Perplexity.
When I ask the DeepSeek Reasoning Model with R1 your exact question, here is the response:
"DeepSeek's knowledge cutoff dates vary by model version, with the latest information indicating:
DeepSeek-R1 Model
Knowledge cutoff: July 2024
DeepSeek-V3 Model
Knowledge cutoff: October 2023 "
So, the model was released in January 2025, but it was trained on information up to July 2024.
When I asked R1 "why" this is, the response was: "The July 2024 cutoff aligns with DeepSeek's development pattern - their DeepSeek-Coder-V2 model (released July 2024) had a November 2023 knowledge cutoff10, showing a typical 5-8 month gap between data cutoff and model release."
Thanks Diana. You keep us well-informed about AI technology. My principal concerns are labor displacement and economic harms. I don't buy into the extinction of mankind hand-wringing, but layoffs are already occurring and the rate will increase in tandem with continuing development of AI and robotics.
Hi Stuart, yes, I agree there will be job displacement. My industry- writers- were one of the first impacted when language models burst on the scene. I chose to reinvent myself as someone who wrote ABOUT language models. It required retraining myself in the technology, which took about a year of taking classes and reading white papers, articles and textbooks. I did it because I loved the subject material, but I also "future proofed" myself. Unless we face a mass-extinction event where knowledge of A.I. is no longer a desirable skill, I gave myself new, more desirable skills in a quickly-changing job market. I was 53-years old when I started this transformation. So, while I am very empathetic to all of those who will be impacted, I also encourage folks to think ahead, plan ahead, and add those awesome new skills before they find themselves in a no-win situation.
Diana, I love the style you write and their structure. Thank you for being so pedagogic.
Thank you! I think this comes from my most recent position working in an elementary school STEM lab. I had to constantly adapt lessons to different age groups and learning styles, (the kids in the "gifted and talented" program needed a different sort of learning approach than the kids in the standard afterschool program.) I took it as a personal challenge to engage all of these groups, all of the time.
When I write AI articles, I am acutely aware I have a very diverse audience. Some of my readers have a very high-level knowledge of AI-terminology. But, some readers are new to AI, but very much want to learn more. I also have readers who speak English As a Second Language. To meet the needs of all these groups, I added the Vocab Key.
In the event my writing was not clear, I added the FAQs. The FAQs reinforce the main points of the article, but worded in a different manner.
In the event someone prefers to listen rather than read, I offer the podcast.
In the event someone is a visual learner, I offer the charts and diagrams.
It takes a lot of energy to do the articles. I always tell my husband each one is like "giving birth." But, I always feel a great sense of accomplishment when I am done.
Thank you for your kind words.
Diana, why does deepseek's data seem end in July 2024?
Hi Stuart!
This is a great question, and it depends upon which model of DeepSeek you are referring to. The easiest way to answer the question is to simply ask DeepSeek. I access DeepSeek through Perplexity.
When I ask the DeepSeek Reasoning Model with R1 your exact question, here is the response:
"DeepSeek's knowledge cutoff dates vary by model version, with the latest information indicating:
DeepSeek-R1 Model
Knowledge cutoff: July 2024
DeepSeek-V3 Model
Knowledge cutoff: October 2023 "
So, the model was released in January 2025, but it was trained on information up to July 2024.
When I asked R1 "why" this is, the response was: "The July 2024 cutoff aligns with DeepSeek's development pattern - their DeepSeek-Coder-V2 model (released July 2024) had a November 2023 knowledge cutoff10, showing a typical 5-8 month gap between data cutoff and model release."
Thanks Diana. You keep us well-informed about AI technology. My principal concerns are labor displacement and economic harms. I don't buy into the extinction of mankind hand-wringing, but layoffs are already occurring and the rate will increase in tandem with continuing development of AI and robotics.
Hi Stuart, yes, I agree there will be job displacement. My industry- writers- were one of the first impacted when language models burst on the scene. I chose to reinvent myself as someone who wrote ABOUT language models. It required retraining myself in the technology, which took about a year of taking classes and reading white papers, articles and textbooks. I did it because I loved the subject material, but I also "future proofed" myself. Unless we face a mass-extinction event where knowledge of A.I. is no longer a desirable skill, I gave myself new, more desirable skills in a quickly-changing job market. I was 53-years old when I started this transformation. So, while I am very empathetic to all of those who will be impacted, I also encourage folks to think ahead, plan ahead, and add those awesome new skills before they find themselves in a no-win situation.