AI race propels Tech firms

Happy, Monday.🫨 As we step into a new week, let's explore AI breakthroughs, advancements and much more.

Here are the 4 interesting AI things that I learned and enjoyed this week.

AI Research

The RPG (Recaption, Plan, Generate) framework is a new approach in text-to-image conversion, excelling in complex prompts involving multiple objects. Traditional models struggle with such complex prompts, particularly with overlapping objects and high training costs. It works by first making the text description more detailed, then dividing the image into different areas for each part of the description, and finally creating each part separately before combining them. RPG outperforms ten other text-to-image models in handling prompt complexity and understanding object relationships. I’m excited to see which text-to-image platform adapts this framework first.

AI Meme
Figure AI, a startup focusing on humanoid robots, is discussing potential funding of $500 million with Microsoft and OpenAI. If they get the money, the company's value could go way up, making it an unicorn startup again. Figure 01, a humanoid robot by Figure AI is designed to handle high-risk tasks and aimed at reducing workforce shortages by taking on challenging jobs. It's like the tech world's robot Olympics – just a month after Tesla shows off its humanoid robot, Optimus Gen 2, big bucks are now being funnelled into Figure AI. This potential deal shows how hot the AI market is right now, with every big player wanting a piece of the action.

AI Concept

In our first edition we set things straight about what artificial intelligence means. In today’s edition we will be answering what is generative ai. Deep Learning, a type of ML, employs neural networks to interpret complex patterns and can use both labeled and unlabelled data. It's divided into generative models, which create new content, and discriminative models, which classify data points. Gen AI falls under deep learning. It uses artificial neural networks to process and learn from data, generating new content such as text, images, or audio from existing content(training). This results in the creation of a statistical model. When given a prompt, Gen AI uses statistical model to predict what an expected response might be. It’s Generative AI if the output is natural and interpretable by humans, like language or visuals.

AI Hacks

This week we are going to see how to make ChatGPT play the role of a data analyst.

1. Create a Chat GPT Plus Account: Go with what you know, like Chat GPT, for ease of use.

2. Explore GPTs: On the left pane, find the 'Explore GPTs' tab, choose the data analyst GPT by Aydin Efendi, and open the chat

3. Select and Download Data: Choose your dataset, download it as a CSV (or other format). Kaggle is a good source for sample data. I used Netflix India Shows & Movies dataset.

4. Upload and Instruct Chat GPT: Upload the CSV in the chat and instruct Chat GPT to clean the data and prepare for analysis. I gave following instruction:

   Data cleaning (release_date field)→ Data analysis(calculate the age) → Visualization (distribution)

5. Ask Questions: Key to good analysis is asking insightful questions. I asked the following question:

   Create a visualization about Aging of shows on Netflix

Below is the result. I see that Netflix India's content strategy, including their focus on refreshing the library with new titles and maintaining a mix of older and newer content to cater to a wide range of viewer preferences.

And that's all I have for you today! GPT store has some noteworthy GPTs that can save a lot of time at work. More about them in next edition. If you find today’s session useful share it with a friend.

⏭️ Stay curious, keep questioning.