xAI, the company owned by Elon Musk, has launched Grok-2 and Grok-2 mini in beta release. This new model is being speculated to have better chat and coding capabilities. But now it can also generate images for you. And looks like it has fewer restrictions than other popular AI Image Generators in the market.
Welcome Grok-2
Grok-2 and Grok-2 Mini are now available in beta release from August 13, 2024. They have better capabilities in chat, coding and reasoning capabilities than the previous Grok 1.5 model.
“Grok-2 is our frontier language model with state-of-the-art reasoning capabilities. This release includes two members of the Grok family: Grok-2 and Grok-2 mini. Both models are now being released to Grok users on the X platform.”
Both are available for beta access to X premium users.
The company has tested an early version of Grok-2 on the LYMYS leaderboard under the name ‘sus-column-r’. They reported that it was outperforming both Claude 3.5 Sonnet and GPT-4 Turbo.
Even in benchmarks like graduate-level science knowledge (GPQA), general knowledge (MMLU, MMLU-Pro), and math competition problems (MATH), the model has performed significantly better than many models available in the market right now.
But that is not what the people are talking about the most. The topic of the town is the now you can generate AI images with it on X.
Grok 2 Can Generate Images on X
On X (Twitter), you can now create images using Grok-2 AI. The model is reportedly using Black Forest Lab’s Flux 1 AI model for this feature.
I am obsessed with new Grok. pic.twitter.com/xZHWif77Nv
— The Redheaded libertarian (@TRHLofficial) August 14, 2024
However, the highlight is that the model does not have strict limits on image generation and no watermarks to prove they are AI-generated as well.
That has led to many people generating somewhat offensive images of famous personalities and politicians. Such images are not allowed in other popular LLMs like ChatGPT.
Grok 2.0 will do political illustrations and real people, while ChatGPT refuses.
— Benjamin De Kraker 🏴☠️ (@BenjaminDEKR) August 14, 2024
This instantly makes Grok 10x more fun…… pic.twitter.com/yDBJO0jWba
Some users are obsessed with the image generation quality and go on to claim that it is the best image generator available in the AI world right now.
Users are happy that it doesn’t have any guardrails and even copyrights. They generated images of cartoon characters like Tom and Jerry, and McDonald’s company and even created a fake Nike shoe page site.
Doesn’t look like there are any restrictions currently on people, brands or copyrighted characters. pic.twitter.com/281XeLXn7u
— CHRIS FIRST (@chrisfirsttt) August 14, 2024
Some of them have created offensive political illustrations like Democratic presidential candidate Kamala Harris holding up a sign that reads “Vote for Trump”.
I asked Grok to create an image that would offend the Democrats. pic.twitter.com/FaTQ81Fqz2
— DogeDesigner (@cb_doge) August 14, 2024
This may help increase the spread of misinformation which is not good especially since the US elections are just around the corner. Many users are also worried about issues with copyrights and the creation of obscene images.
Many more examples include cartoon characters in gory backgrounds, celebrities holding weapons, politicians doing illegal things and some NSFW images. While this shows that the model is more free, it also makes it vulnerable to many other issues.
The Verge has found that the model itself states its limitations it will not create images that infringe on existing copyrights or images that can harm others including deep fakes, it looks like that is not what is happening actually.
Conclusion
The new image generator is awesome but Grok 2 is powerful and the image generation quality is fantastic but the potential misuse for the company as authorities will be looking more into regulating AI.