Justin's AI Newsletter

The most interesting content, news, and insights in AI each week.

--

Hey friends! This week I created a content library featuring some of my best videos and writing about AI. I realized that I've been producing a lot of content and it may be difficult for people to keep track of it all. So, I put everything in one place.

If you enjoy my content, please feel free to share it with a friend or colleague. I created this library of AI content to make it easy for people to find interesting things and learn more about what they want in AI.

👁️ AI Agents and a Peek into the Future of Goal-Driven Intelligence

This week, the AI community has been excitedly talking about the concept of "AI Agents." These experimental software programs are grabbing attention for their ability to come up with their own prompts based on a given goal.

Imagine a world where you don't give ChatGPT specific tasks, but instead, you provide it with overall goals. You wouldn't say something like "write me an SEO blog post," but rather "improve the overall conversion rate of our business." The AI would then continuously come up with tasks that would move towards that goal, and execute them autonomously.

We've already seen some amazing progress in AI with ChatGPT, which can understand and generate ideas, and ChatGPT plugins that allow AI to perform various actions. But with AI agents, we'll see AI systems that not only figure out what needs to be done to reach a goal, but also take action to get it done.

As we continue to explore the potential of these AI agents, we may find ourselves on the verge of a new era in artificial intelligence—one that is goal-driven and increasingly autonomous.

Some more resources around AI Agents:

• If you know how to code, you can experiment with AutoGPT, one of the first tools released, here.
• Someone built AgentGPT which allows you to create your own AI agents (not perfect, but fun to experiment with).
• ​HyperWrite announced their AI Agent this week. In the demo, Matt asked the AI to order a pizza, and the assistant went through the entire ordering process.
• Lastly, here is a demo of me using the HyperWrite agent to subscribe to this newsletter...

Justin Fineberg @JustinFineberg Playing around with @HyperWriteAI’s new Assistant… I got the assistant to look me up, subscribe to my newsletter, and summarize the latest issue. All in a single prompt. pic.twitter.com/XDgAxyYemn April 14th 2023 3 Retweets 27 Likes

🚨 Is your data safe in ChatGPT?

I’ve been getting asked a lot about privacy and data with ChatGPT, so let’s break it all down.

First off, ChatGPT can use the data you input into ChatGPT for future model training. Your conversations can be reviewed by OpenAI and trainers. However, they make it clear that they remove all personally identifiable information from the data when they use it to improve the model.

But, you actually have the option to opt-out of having your data used by OpenAI by filling out this Google Form from this page here.​

Personally, I'm not too concerned about this issue. I don’t have too much information that is so private that I would worry about it going into the massive dataset for training the model, and I don't think most individuals need to worry about it. However, businesses do have the option to opt-out of model training or use the API, which doesn't use their data to train the model.

This is all to say that as AI technology advances, OpenAI is likely to adapt in response to the mounting demand for stricter data policies and enhanced transparency. Given the increased reliance of businesses on ChatGPT and the escalating pressures from governments to impose more rigorous data regulations, OpenAI is set to confront these challenges by establishing more exhaustive and transparent practices.

​

💡 This Week's Recommendations

​Biden Administration Weighs Possible Rules for AI Tools Like ChatGPT: The Biden administration is considering whether to regulate artificial intelligence tools like ChatGPT due to concerns about discrimination and the spread of harmful information. The Commerce Department has requested public comment on accountability measures, including whether new AI models should be certified before release. This is just the start of government regulation in the AI game.

​Amazon enters the generative AI race: Amazon has launched Amazon Bedrock, a service that enables customers to build generative AI-powered applications using pretrained models from third-party startups such as AI21 Labs, Anthropic, and Stability AI. This service is designed for large customers who are building "enterprise-scale" AI applications. It represents Amazon's biggest AI push to date, and a significant bet that businesses are seeking access to customizable models that do not require billions of dollars to develop.

​Building LLM applications for production: An extremely in-depth look at building large language models for use in products. It is essentially the LLM textbook, so I wouldn’t suggest reading the whole thing 😴 . However, if you are considering incorporating AI into your business, you may want to pass this along to your engineers…

Thanks for tuning in,

Justin

​

​Justin's Full Content Library | Previous Newsletters​

113 Cherry St #92768, Seattle, WA 98104-2205
​Unsubscribe · Preferences​