Artificial Mind Chatbots Compete for Dominance in these AI Wars

ChatGPT was released on November 30, 2022. To use it, we need to create an account with our name, email address, and phone number. In less than two months, ChatGPT had more than 100 million registered users, making it the most popular website of all time:

‍

"ChatGPT, the popular chatbot from OpenAI, is projected to have reached 100 million monthly active users in January, just two months after it was released. This makes it the fastest-growing consumer application in history, according to a UBS study released on Wednesday."

‍

ChatGPT was able to grow so quickly because of word-of-mouth between friends, mentions on social media, and a lot of media coverage.

‍

This is just one tiny example of how much ChatGPT has changed the Internet. ChatGPT has redirected some of the largest companies in the world in just two months, kudos to its extensive capabilities and popularity.

‍

Microsoft Makes The First Move

Microsoft became one of the first tech giants to make a public move after ChatGPT gained popularity. Microsoft spent $10 billion on OpenAI, the developer of ChatGPT.

‍

Reportedly, Microsoft intends to spend $10 billion on ChatGPT, the developer of the popular artificial intelligence (AI) platform.

‍

Then, a few weeks later, Microsoft announced that it will incorporate ChatGPT's functions into its Bing search engine and Edge web browser.

‍

He begins by stating, "Bing's integration of artificial intelligence will fundamentally alter the search experience." The demonstration then illustrates his point. Towards the conclusion, he is very forthright about one objective: he wants Bing to steal some market share from Google's search engine supremacy. 

‍

Google Announces Bard And Several Other Things

Google examined ChatGPT and issued a red alert back in December:

‍

According to reports, Google's management has declared a 'code red' in response to the growing popularity of the ChatGPT AI.

‍

"Sundar Pichai, the CEO of Google and its parent company, Alphabet, has attended several meetings regarding Google's AI strategy and urged numerous groups inside the firm to refocus their efforts on countering the danger posed by ChatGPT to Google's search-engine business... The Times said that teams in Google's research, trust, and safety divisions, among others, have been tasked with assisting in the creation and deployment of AI prototypes and products.

‍

A month later, Google's CEO Sundar Pichai announced the launch of Bard, an AI conversation tool.

‍

"Two years ago, we introduced cutting-edge language and conversation capabilities enabled by our Language Model for Dialogue Applications" (or LaMDA for short). We have been developing an experimental LaMDA-powered conversational AI service called Bard. Today, we're taking another step forward by making it available to trustworthy testers prior to releasing it to the general public in the coming weeks. Bard aims to integrate the breadth of global knowledge with the strength, intelligence, and creativity of our massive language models."

‍

You may recall that in June of last year, Blake Lemoine, a Google employee, declared that he believed LaMDA to be conscious.

‍

LaMDA was never really able to be seen by the "general public" as a whole. This was more of an experimental system that was intended to be used within the Google network. There will be something of a coming-out party for LaMDA at the Bard.

‍

Bard is not the only thing Google has done for Bard. There is a company called DeepMind that Google owns, which was acquired by the company in 2014. Sparrow is a new AI product developed by DeepMind:

‍

We'll be learning everything you need to know about Google's upcoming ChatGPT killer AI 'Sparrow' soon. A chatbot called Med-PaLM was also announced by DeepMind in January, which was designed to help doctors. There is no doubt that its performance is impressive:

‍

Earlier this week, Google and DeepMind launched MedPaLM, a large language model aligned with the domain of medicine

‍

“Based on the results of a study conducted by a group of healthcare professionals, it was determined that 92.6 percent of the answers provided by Med-PaLM were on par with the ones provided by clinicians (92.9 percent).”

‍

In addition to Google, there are also hundreds of millions of dollars invested in another company named Antropic, which has its own chatbot named Claude.

‍

Anthropic, a rival to ChatGPT, receives a $400 million investment from Google

‍

“Founded in 2021 by former leaders of OpenAI, including siblings Daniela and Dario Amodei, Anthropic AI in January released a limited test of a new chatbot named Claude to rival to OpenAI’s wildly popular ChatGPT.”

‍

In 2023, it will be fascinating to see how all of this work in Google's favor in terms of LaMDA, Bard, Sparrow, Med-PaLM, and Claude.

‍

What About China?

Among the most popular search engines in China, Baidu has become the dominant one. To make sure they do not fall behind, Baidu has announced Ernie Bot ("Enhanced Representation through Knowledge Integration") to compete with Google:

‍

It comes as Baidu announces plans to launch a ChatGPT-style 'Ernie Bot' in the near future, which has now risen to an 11-month high

‍

“As time has progressed, the system has evolved into a series of advanced big models that are capable of handling a wide range of tasks, including language understanding, language generation, and text-to-image generation.”

‍

The last part of the sentence is quite interesting because Ernie Bot seems to be a combination of text generation and image generation in one system. A Chinese system called Wu Dao 2.0 was one of the first to combine these two techniques and also to use massive amounts of training data in order to achieve results in 2021:

‍

The GPT-3 scared you? You should meet Wu Dao 2.0: The monster with 1.75 trillion parameters

‍

"Wu Dao, which means Enlightenment, is another language model that is similar to GPT in terms of its structure. Jack Clark, OpenAI's policy director, calls this trend of copying GPT-3 "model diffusion." Yet, Wu Dao 2.0 holds the record for having the most parameters among all the copies (10x GPT-3). Coco Feng, a reporter for the South China Morning Post, reported that Wu Dao 2.0 was trained on 4.9TB of high-quality text and image files, making GPT-3's training dataset (570GB) pale in comparison. In spite of this, it is worth noting that OpenAI researchers curated 45 TB of data to extract clean that 570 GB.” 

‍

There is no doubt that the release of these massive combined models now should shake things up even more now than before.

‍

What About Apple And Facebook (Meta)?

Before ChatGPT was released, the parent company of Facebook, Meta, had released a public chatbot. Galactica is the name of the game. As soon as Galactica became private again, it almost immediately became a public company again. Due to the fact that Galactica focuses on research papers, it has a smaller amount of training data of higher quality, as a result. What is the reason for the removal of the website? 

‍

Galactica, Meta's new large language model, was taken down three days after it was launched

‍

Last Tuesday, Meta's much anticipated artificial intelligence model Galactica was unveiled to the public for the first time as a demo at its annual conference. However, before the weekend, the tool was removed from the site as it had a tendency to produce gibberish and was incapable of solving some rudimentary mathematical questions with accuracy, resulting in its removal from the site before the weekend. A new initiative from Meta, Galactica, aims at streamlining scientific research discovery and easing the creation of scientific texts. The Meta team described it in a paper as a tool for organizing science. They also said it was trained on 48 million papers, textbooks, lecture notes, millions of compounds, proteins, scientific websites, and encyclopedias which total 120 billion parameters.”

‍

There seems to be a lot of bad press surrounding Galactica because it wasn't ready when it was released.

‍

In the chips that power its smartphones, Apple incorporates a "neural engine" that uses AI. In order to speed up the processing of AI, the neural engine is used. There is, however, no chatbot on the horizon as of yet for Apple, and it could take some time until they do. 

‍

Open Source Competition To ChatGPT

In the above description of everything, one of the problems is the fact that it is all “Big Tech”. As a result, there is sometimes a tendency for Big Tech to play things safe. The Bing chatbot, for instance, has been reported to have not been able to write cover letters for people applying for jobs according to a report from one early user of the chatbot.

‍

ChatGPT is more than happy to create cover letters for you, but for how long? Several things that ChatGPT would do in December that it will not do today are things that it would do in December. ChatGPT seems to be getting more restrictive with every passing moment, and it seems like it's only going to get worse.

‍

The other is called BLOOM:

‍

“A significant impact has been made on AI research due to large language models (LLMs). From a user's instructions, these powerful, general models can take on a wide variety of new language tasks based on a variety of language tasks. There is, however, a significant challenge facing academics, nonprofits, and smaller companies' research labs that want to create, study, or even use LLMs, as only a few industrial labs with the required resources and exclusive rights are able to do so. As a result of the largest collaboration of AI researchers ever collaborated on a single project, we are releasing BLOOM today, the first multilingual LLM trained within complete transparency, as a response to this status quo. BLOOM is able to generate text in 46 natural languages and 13 programming languages thanks to its 176 billion parameters. BLOOM will be the very first language model with more than 100B parameters created in history for almost all of them, such as Spanish, French, and Arabic. There has been a year of work involving over 1000 researchers, from over 70 countries, and from over 250 institutions, culminating in a final run of 117 days (March 11 - July 6) on the Jean Zay supercomputer in the south of Paris, France, where the model was trained over the course of 117 days (March 11 - July 6), thanks to a computational grant from the French research agencies CNRS and GENCI worth €3 million.”

‍

There is currently no comparison between BLOOM and ChatGPT when it comes to the quality of their responses, but it may only be a matter of time before the two reach the same level of quality.

‍

Where Did All These Chatbots Come From?

This article has discussed a variety of systems:

• GPT-3
• ChatGPT
• Bing
• LaMDA
• Bard
• Sparrow
• Med-PaLM
• Claude
• Ernie Bot
• Mu Dao 2.0
• Galactica
• Open Assistant
• BLOOM

‍

The way they work is about the same for all of them. There are a lot of AI innovations happening at the same time, so how can that possibly be possible? The reason for this is due to the fact that a scientific paper published in 2017 outlined the main technology that all of them were based on:

‍

Then each vendor can add to or customize the data.

‍

This is essentially what GPT-3 is all about. In order to train the tool on the Common Crawl or some similar tool, they took the paper "Attention is all you need".

‍

The innovative feature of ChatGPT is that they have taken GPT-3.5, and then they trained it to behave more as a human would. There are thousands of sample queries that can be given to the model to produce responses that a human might want to receive based on the queries that a human might ask. As a result of this, ChatGPT sounds a lot more human-like. Our users of ChatGPT today have been used to gather data and help refine the model even further using the collective wisdom of all of us - all of the ChatGPT users today.

‍

In any case, the only real barrier to creating these chatbots would be the amount of computational power needed to train the model and then run that model once it has been trained. This computation can cost upwards of a million dollars if it is done correctly.

‍

Due to the fact that nothing that ChatGPT does is really secret or hidden, it is easy for lots of people to copy what ChatGPT does.  By the end of 2023, there will be a lot of chatbots competing for our attention.

‍

Who Is Going To Win?

We asked ChatGPT: “Tell me a story about ChatGPT beating out Google.”

‍

Here is the story that ChatGPT created:

‍

'Once upon a time, in a world dominated by technology, there lived a language model known as ChatGPT which existed in a world where technology dominated nearly everything. A company called OpenAI, which is known for developing cutting-edge artificial intelligence systems as well as ChatGPT, created the application.

‍

One day, the announcement was made that there would be a competition in which the best search engine would be crowned the winner. Despite the fact that it was the biggest search engine in the world, Google was confident that it would easily win. Despite this, ChatGPT was determined to give it a good run for its money, so it decided to compete with it.

‍

In order to provide answers to search queries in a matter of seconds, Google's algorithms scanned through millions of pages within the first few seconds of the competition starting. ChatGPT, on the other hand, used its advanced language processing skills to understand the context and intent of each query via its advanced language processing skills. As a result, it was able to provide highly relevant and accurate answers based on this understanding.

‍

During the course of the competition, people began to notice the capabilities of ChatGPT as the competition progressed. It was not only faster than Google to reply, but it also provided more insightful and personalized responses than Google did. In addition to having answers to every question, ChatGPT also had so good answers that people were amazed at how good their answers were.

‍

It didn't take long for ChatGPT's popularity to skyrocket, and it soon became the go-to search engine for millions of people around the world. There was no doubt in my mind that ChatGPT had won the competition, as Google was left in the dust.

‍

As the news spread like wildfire, ChatGPT and its incredible capabilities soon became the talk of everyone who was interested in technology. It was said that OpenAI was the company that had created the future of search engines, and ChatGPT became a household name as a result.

‍

Thereafter, ChatGPT continued to improve and innovate, and it remained the top search engine for a number of years to come as a result. As a result, ChatGPT was able to exceed the brilliance of Google, while Google had to play catch up to keep up.

‍

That, my friend, is how ChatGPT beat out Google in the search engine battle and became the king of search engines.”

‍

A future like this is interesting to ponder, and it is worth considering.