“A year spent in artificial intelligence is enough to make one believe in God.” - Alan Perlis

Monday’s email was pushed to today, March 7th 2023

In today’s email:

• News: Facebook’s LLaMA model leaked for pubic use

• Zoom out: Quantifying ChatGPT’s value

• New projects: AI teaches itself APIs

• New tools: Ariana: Your ChatGPT assistant in Whatsapp

Researchers plan to create biocomputers powered by human brain cells

Artificial Intelligence (AI) has come a long way, but many ways it still pales in comparison to the power of the human brain.

A new approach called organoid intelligence (OI) is now being developed, where lab-grown brain organoids serve as biological hardware. This interdisciplinary field has the potential to revolutionize computing and pave the way for a new era of fast, powerful, and efficient biocomputing.

What are brain organoids?

Brain organoids are lab-grown cell cultures that share key aspects of brain function and structure. Although they are not "mini-brains," they contain neurons and other brain cells essential for cognitive functions such as learning and memory. Organoids have a three-dimensional structure, which increases the culture's cell density 1,000-fold and allows neurons to form many more connections.

To make OI a reality, brain organoids need to be scaled up to around 10 million cells. Scientists are also developing technologies to communicate with organoids and send them information and read out what they are "thinking."

Researchers are adapting tools from various scientific disciplines, such as bioengineering and machine learning, and engineering new stimulation and recording devices.

OI is not only revolutionary for computing but also has the potential to help prevent and treat neurological conditions. Brain organoids can be produced from adult tissues, and scientists can develop personalized organoids from skin samples of patients suffering from neural disorders such as Alzheimer's disease. They can then investigate how genetic factors, medicines, and toxins influence these conditions.

Creating human brain organoids that can learn, remember, and interact with their environment raises complex ethical questions. To address these issues, the authors have partnered with ethicists to establish an "embedded ethics" approach. All ethical issues will be continuously assessed by teams made up of scientists, ethicists, and the public as the research evolves.

While OI is still in its infancy, a recently published study provides proof of concept. A team showed that a normal flat brain cell culture can learn to play the video game Pong. From here on, it's just a matter of building the community, the tools, and the technologies to realize OI's full potential.

Quantifying ChatGPT’s labor contribution to society

The ChatGPT API was recently released at a cost of 500,000 tokens per dollar.

What is a token?

In the context of ChatGPT, a token refers to a sequence of characters that represents a unit of meaning. In natural language processing and machine learning, tokens are often created by breaking down a piece of text into smaller units, which are usually words, but sometimes simpler subwords or individual characters. Each token is then assigned a numerical value, which can be used by machine learning models to analyze and process the text.

For example, when processing a sentence, ChatGPT will first tokenize it by breaking it down into individual words, punctuation marks, and other meaningful units before generating a response.

So we have a machine capable of generating a half a million of these tokens for about a dollar. How do we quantify that contribution to society?

The intelligence of ChatGPT is difficult to measure using traditional means such as IQ, though estimates have ranged from 83 to 147.

Despite this, ChatGPT is capable of communicating in thousands of languages, including Egyptian Hieroglyphics and Sumerian Cuneiform. It has been exposed to about 50Tb of written text, which is a sizable amount of all the existing words on the internet.

In terms of intelligence, ChatGPT could be classified as having an apprentice-level of intelligence, being able to perform well on highly specific tasks, but not others.

Although it would not be advisable to let ChatGPT perform tasks such as designing a car or arguing a case in court, it would be able to provide advice on these matters in situations where a professional is not available.

Consider this: it has been estimated that a human can think at a rate of roughly 600 words per minute, with each word costing approximately 1.3 tokens. Therefore, a human working at maximum output for 40 hours per week for a year could produce approximately 97 million tokens of thought, which would cost $195 out of ChatGPT, roughly $0.09 per hour.

The monetary value of human thought raises questions about the value of intellectual labor in comparison to physical labor and other forms of work.

The shift towards knowledge work in the Western world has been a significant development over the past 40 years.

But if AI technology continues to advance and automate many of the tasks traditionally performed by humans, there is a growing concern that this trend may be reversed.

In other words, rather than continuing to push for more knowledge work, society may instead find itself in a situation where humans are forced to focus on the aspects of work that cannot be replaced by AI.

This could lead to a renewed emphasis on creativity, emotional intelligence, and other human qualities that cannot be replicated by machines.

SLAPA: the Self-Learning Agent for Performing APIs.

SLAPA is a developing program that is capable of teaching itself how to use tools, specifically APIs, in real-time, completely automated.

One of the most impressive things about SLAPA is the way its knowledge accumulates over time:

As more people use the program, its knowledge base becomes more comprehensive, meaning that it can provide more accurate and efficient responses to user queries.

SLAPA gets better at writing API calls the more it is used, making it an increasingly powerful tool for developers and other users.

When given a task, SLAPA knows to search for API documentation and learn all the necessary information. If the initial API calls don't work, SLAPA learns from its mistakes and tries again until it succeeds.

This ability to learn from experience is what sets SLAPA apart from other AI agents. By being able to teach itself how to use tools, it can adapt to new situations and tasks in real-time, making it an incredibly invaluable tool for anyone working with APIs.

Ariana - The ChatGPT assistant on Whatsapp

Enter Ariana, the ChatGPT-powered Assistant that lives within WhatsApp and helps you get answers on the go.

Ariana is a portal to ChatGPT living in your WhatsApp workspace, designed to help you with day-to-day tasks.

You can ask her anything from A-levels questions to healthy diets or work-related queries, and she will respond in just a few seconds.

To use Ariana, you need to add her as a contact on your WhatsApp.

Your questions and conversations with Ariana are completely private and only accessible to you and Ariana. This means that only you will be able to see the questions and answers that you ask her directly.

Ariana is like having a personal assistant at your fingertips, available to help you whenever and wherever you need her. Whether you're a student, a professional, or just someone who needs a quick answer to a question, Ariana is the perfect solution.

^{sponsored post}

Stay up to date on all things AI

Thank you for being one of the 15,000+ new readers since the beginning of the year. Your readership means the world to us and this Monday morning email will always be free.

AI Breakfast also offers a premium option:

As a Premium Member for $97/yr, you get:

-Bonus Wednesday Email
-Bonus Friday Email
-Discounts on AI tools (like Jasper AI)
-Special discounts on AI conferences
-FREE digital copy of our upcoming book, “Decoding AI: A Non-Technical Explanation of Artificial Intelligence” Available April 18th
- Never miss an update in AI.

Special offer: 55% off to new subscribers with promo code “PAPERCLIPS”
(Before you click: If you’re on mobile, you can enter the promo code from the ‘details’ tab in the payment link)

Thank you for reading!