Welcome, superheroes!

This week, we want to show you how to save money while using AI. Feel free to give us feedback with a rating at the end of this issue.

Here’s what we have on the menu today:

• Top 5 free chatbot platforms

• The AI chip race

• Let’s talk open source

• Google Gemini battles GPT-4

• A new image generator battles DALL-E

Enjoy!

🛠️ AI Tools - Sneak Peek

Freedom of Creativity with Chatbot Platforms

In today’s issue, we check alternative generative AI platforms that can be used for free: either open source or with a free plan.

HuggingChat, Replicate, PartyRock AWS, Gemini, and Claude offer a spectrum from open-source projects to cloud services, making chatbot platforms more accessible.

This Friday, you can learn more about HuggingChat. Stay tuned!

📈 AI Trends

The AI Chip Race

ChatGPT alone requires thousands of Nvidia’s finest chips. With the rise of generative AI, the demand for microchips is strongly outweighing supply.

This trend sparks a chip race among tech giants like Nvidia, AMD and Intel, urging them to expand their offerings.

With a market worth billions, other big tech players like Microsoft, Google, Amazon, Meta, and OpenAI are starting to work on their own AI chips.

We see the demand for chips only increasing as more powerful AI models appear on a regular basis and AI applications are spreading throughout all industries.

Do we always need to use the most powerful models to help us? Keep reading to find out. 👇

👨‍🏫 Learn with Us

Introducing Open Source

Large language models from big companies like OpenAI (ChatGPT), Google (Gemini), Meta (Llama family), and Anthropic (Claude) dominate media reports.

In parallel, open source models are on the rise. These are trained AI models, which you can freely download and use.

🏢 Platforms

Hugging Face. The biggest home for open AI models: language, image, and video generation engines, including open ones from Meta and Google.

Civitai focuses on image and video generation, with a visual interface to help you find the right models and extensions for your purpose.

Such platforms have strong communities, offering:

• great tutorials,

• active discussion threads,

• often playgrounds to try models online.

For serious use, open source models need:

• to be downloaded (often several GB),

• used in a compatible software framework,

• run on your PC (strong hardware needed),

• or run on a cloud computing service.

Increasingly, smaller models are appearing. They can be run on office laptops and even smartphones.

🚀 How to start

It’s not that complicated. You can use the web platforms mentioned in AI tools: HuggingChat and Replicate. No download or set-up needed.

📚 Learning material

Want to learn how to use Hugging Face for free? Then, we recommend DeepLearning.AI - Open Source Models with Hugging Face, from the AI legend, Andrew Ng.

The course covers how to use Hugging Face for just about any currently popular application.

📌 In summary

Is open source the future of AI? Maybe. In the past, open software led to more transparency, high quality code, and constant improvement via crowd intelligence.

✍️ Write Like a Pro

Is ChatGPT Plus the Best?

Ever struggled writing compelling social media posts? We’ve been there. Doing research and editing can be a pain.

GPT-4 can do that if you have the ChatGPT Plus plan, but is it worth the price? We decided to put it to the test against Google’s Gemini (ex Bard).

Here’s the prompt we used:

You're an expert LinkedIn copywriter and influencer. I want you to write a LinkedIn post about the top 5 current trends in AI. Your target audience are non-technical AI enthusiasts. Use clear and conversational tone. The post should have a one-liner hook. Use bullet points to list the trends. Make sure each line is less than 65 characters. Add a call to action one-liner at the end of the post. Use web-search and list source URLs at the end of the post. Avoid using emojis. The post must be 100 - 120 words.

And here’s what Gemini and ChatGPT-4 produced.

ChatGPT-4

🔴 Slow
🟡 Web search
🔴 Language quality
🟡 Prompt adherence
🟡 One result by default

Google Gemini 🏆

🟢 Fast
🟢 Web search
🔴 Language quality
🟡 Prompt adherence
🟢 Three results by default

Looks like you don’t always need to pay for ChatGPT Plus. Still, both posts need work to make them more personalized.

Once you have the copy down, it’s time to create some images. Keep reading to check out our new image generator battle. ⬇

👨‍🎨 Visual Experiments

Image Generator Battle: DALL-E vs Ideogram

This Tuesday, we’re demonstrating smart watch prototype design. DALL-E will compete with Ideogram. Let’s take a look. 👇

So let’s compare, shall we?

DALL-E

🟢 Prompt adherence
🟡 Visual appeal
🟢 Realism

Ideogram 🏆

🟢 Prompt adherence
🟢 Visual appeal
🟢 Realism

Well it looks like Ideogram is taking this one.

Try it yourself → Find your winner.

Here’s the image prompt we used:

Prototype design, photorealistic style, high-tech wraparound futuristic watch. The watch has a modern and slick feel, featuring neon green elements that stand out against a sleek, futuristic design. The wraparound style is innovative, with a curved screen that conforms to the wrist. The watch interface displays advanced functionalities like health monitoring and communication.

With its generous free plan, Ideogram is the new kid on the block of image generators. Check it out here.

🎁 Free Resources

As our subscriber, you have free lifetime access to exclusive resources.

🔬 AI content systems lab — Exclusive content creation GPTs.*

📚 AI resource bank — Ultimate DALL-E ebook, cheat sheets, and guides.

*ChatGPT Plus or Team subscription required.

How was this issue?