The AI Impact Big 3
Meeting at the intersection of AI and practical use cases
Welcome again to the new AI Impact newsletter and thanks for signing up!
This is the Big 3 - a look at the biggest topics in AI this week and why it affects your business.
This Week: What Does GPT Even Mean? Amazon Steps Up Its AI Game, and of course….One Big Thing! (Do You Have An 🍎? )
One Big Thing: Last Week Meta, This Week Apple 🍎
Apple entered the AI arena last week with its own artificial intelligence tools, including a conversational AI chatbot named "Apple GPT," to compete with leading AI chatbots.
This is another challenger to ChatGPT, the popular LLM released from OpenAI earlier this year.
As we mentioned last week, LLMs are capable of generating text, writing different kinds of creative content, and answering your questions in an informative way.
Why does this matter?
Apple is still far behind the bot competition, but they have the software, hardware, ecosystem, and team to be a massive competitor to existing chatbots.
Other key points:
Apple has developed an internal framework and has deployed a chatbot dubbed “Apple GPT” within internal systems.
Apple GPT will compete directly against leading AI chatbots such as ChatGPT and Bard.
A consumer product is aimed for next year.
Just another example of the huge investment happening in AI research right now.
My Two Cents: All the major players are jumping on board, including what Microsoft has done as we covered last week. AI is quickly becoming more accessible, and we’ll be talking about these real-time use cases moving forward.
2. Amazon…It Was Only A Matter of Time
Amazon is investing more in AI. AWS announced major updates and new services to strengthen its position as the cloud provider of choice for organizations using generative AI.
The announcements include:
Expanding the AI foundation model service called Bedrock to support more models like Cohere and Anthropic Claude 2.
A simple generative AI application to develop without writing any code.
Generative AI was integrated into the Amazon Quicksight business intelligence (BI) service, enabling business analysts to use natural language queries.
My Two Cents: If you haven’t used Claude 2 yet or played around with Stable Diffusion, do it. It is amazing what text, images, etc you can generate. And I have Claude on my phone to be able to ask questions on the go.
3. What Does GPT Mean in ChatGPT?
I like to use this third slot to educate in some way about AI and/or how it’s working inside organizations.
So what do these terms mean? Let’s start with ChatGPT - you’ve heard a lot about this technology in 2023.
There was a great explanation this week from Conversational AI expert Kerry Robinson. Let’s break it down:
G is for Generative (AI)
This is a Language Model. “You can make a language model 'Generative' by giving it a sentence, and asking it for the next word. Then giving it the sentence plus the word it came up with, and asking for the next one, and so on. That's Generative AI. And that's the G in GPT.”
P is for Pretrained (Language Model)
“It takes a lot of data, computing power, time and money to train a Large Language Model so organizations tend to do it only occasionally. And then, because these large models are so powerful, they can use this 'pre-trained' model to do all sorts of different things. These models are so big, and so flexible, they can be pre-trained, and used for lots of different tasks. That's the P in GPT.”
T is for Transformer
A transformer is the name given to a special type of neural network. There are billions of connections in these networks holding a ton of information. “That's why Large, Pre-trained language models can do so many things so well: they encode a lot of knowledge and reasoning power in those billions of connections. Just like our brains do.”
My Two Cents: It’s amazing what GPTs can do and it will be awesome to see how this evolves, in relatively short order. What strikes me is the role the Human still plays in the entire process. Human Feedback is crucial to rating and validating the output of these models, and in turn is used to “reinforce” these models to make them more accurate.
Talk next week and thanks,
Brad
🙏 Thank you for reading AI Impact. Follow me on LinkedIn for more updates on AI and the practical implementation of it, and be sure to refer friends using the link below.