What is Grok AI

Our friend Elon Musk's Grok AI platform has been in the media due to some recent updates in version 2. In addition, we are currently experiencing a period of problems between Platform X and the Ministry of Supreme Court, generating a major controversy in Brazil.

Well, let's delve a little deeper and understand why all this recent media coverage, looking at Grok, because the X Controversy is for a future article.

What is Grok

Grok is an Artificial Intelligence platform developed by the company xAI, Uncle Elon's famous X Group. The first version was released in November 2023. The tool was developed integrated with X (formerly Twitter). As for its database, it was initially thought that they would only use X's own database, which wouldn't work very well, but its database includes information collected from the internet, documents, books, images and various sources that are public. According to xAI, their goal is to have a model that understands the “true nature of the universe” 👀

Grok Features

Grok, as I said before, is integrated with the X platform, it allows users to interact directly through the social network, this allows responding to direct messages, like a chatbot, and even creating content for posts.

Grok’s mathematical reasoning is one of the best and most accurate compared to other AI models. This makes it very useful for complex calculations and mathematical problems.

Starting in August 2024, with the launch of version Grok 2, the tool gains the ability to generate images, which has been causing a stir these days, making the tool even more versatile for applications involving visual analysis and creation of graphic content.

Grok 2 also comes with a processing power of up to 128,000 tokens, allowing for deeper and more complex discussions. Compared to ChatGPT, the standard version has a limit of 8,000 tokens. The latest paid version has the capacity to process up to 128,000 tokens.

And before you ask me what a Token is, here's the thread:

“In simple terms, a token can be a whole word, a part of a word, or even a punctuation symbol. For example, the word “dog” is counted as a single token, while the phrase “I don’t know.” could be split into three tokens: “I don’t,” “I know,” and the period.” ChatGPT

The greater the number of tokens a model can process, the longer and more complex the text it can understand and respond to in a single interaction.

Grok vs ChatGPT

ChatGPT is designed for a wide range of applications, across many industries (education, customer support, software development, and so on…), while Grok is focused on X users and tasks involving mathematics and now images.

The training of both platforms uses a wide range of data, as we have already mentioned. However, Grok was optimized to respond within the environment of platform X, which ends up providing contextualized responses specifically for the social network.

Grok does not have a free version, which makes us very sad. Unlike ChatGPT and several other tools, we can use the free version and it already works very well for us. Currently, you need to have a plan with Grok, which currently costs a monthly fee, as shown in the image below. There is also an annual option, which they say is more advantageous, but you pay all at once. The subscription is also linked to platform X; you do not subscribe only to Grok, as shown in the image.

Values as of August 28, 2024

We recently had an update to ChatGPT that integrated the Dall-E tool into it, which allows us to generate images by requesting them directly through it. The version was also released for free access (although very limited). Grok 2 also has built-in functionality that helps users of the social network quickly create their images.

Quick summary: Grok is ideal for those who use X and need support for complex calculations, free image generation (we'll get to that later) and extensive dialogues. ChatGPT is more generalist, adapted to a wide range of tasks and sectors, with a greater emphasis on textual and conversational support.

Generate images in Grok 2

This is the feature that has been causing a stir. I've been following a lot of things these days about what it can do, and once again, Uncle Musk has created or caused problems. I believe he's a fan of the phrase: Say good things or bad things, but talk about me... lol

I even asked ChatGPT a question asking who this famous phrase is attributed to:

“It is popularly attributed to the Irish writer Oscar Wilde, known for his witty works and aphorisms. Although this phrase captures Wilde's ironic and provocative style well, there is no clear source that proves that he actually said or wrote it. The quote reflects the idea that, in certain contexts, publicity, even negative publicity, can be advantageous, keeping a person or brand in the spotlight.”

The truth is that the generation of images in Grok 2 has generated a lot of controversy these days, due to ethical and technical issues. Here are the main points of this controversy:

Data Usage: The main issue revolves around the way Grok 2 was trained to generate images. There is a large and very concerning discussion about the use of copyrighted data or the lack of such consent from the creators whose works may have been used to train the model.

Quality and accuracy: While some praise the quality of the images generated in Grok 2, others criticize the lack of accuracy in visual representations or the risk of reinforcing stereotypes and prejudices. The tool can replicate some biases that were imputed in the training data.

Regulation and ethics: The controversy involves discussions about the need to regulate the use of AI to generate images. In Grok 2 you can use what they call fun mode and I accompanied my friend Caio Camargo these days creating some images and talking about the new version. The current version allows you to create images representing brands and even exploring themes such as sensuality and violence. If you try it in Dall-E and other tools you will be blocked, they restrict it a lot.

“Although this freedom adds fun, it brings serious risks. Brands can be used without control, violating copyright and damaging their images, while the possibility of creating and disseminating false information becomes even more present.” Caio Camargo - in a post on his Instagram.

Here we return to talk about the ethical part, which is the great controversy of the platform. How far can we go with this “creativity”? I will leave some images here generated by Caião.

And if you do some google searches, you will find images like these below.

Grok in Brazil

Update on 08/30/2024.

Unfortunately, in Brazil, we are experiencing a problem where Platform X has been banned for users. By a single decision, X was disabled and thus users will not have access to Grok. Another issue is that it was also decided that all VPN access applications would be removed from the Apple Store and Play Store platforms so that no user can access Platform X. And just when we thought it couldn't get any worse, we have more information: a fine of 50 thousand reais per day was set for anyone accessing Platform X via VPN access, both companies and individuals.

The measure also blocked the bank accounts of Space X, Uncle Musk's company that distributes internet via satellite. The problem is that the Armed Forces are one of the main users of the service.

Well, unfortunately, at the moment we are experiencing a limitation on access to the platform. As soon as we have a new update, I will bring it here again. This is yet another big controversy that we are experiencing.

Updated on 08/31/2024

The minister backed down and suspended the order that bars the use of VPN (virtual private network) after the decision to suspend X in Brazil. The minister suspended the execution of item 2 of the decision this Friday (30). Moraes had ordered virtual stores to make it difficult to access such applications.

What will be the next steps and news?