AI explains why they’re so bad at drawing human fingers

No matter where you sit on enormous and complex debate regarding AI models and their use of data sets to create new images, artwork, text, and even movies., there’s one thing that everyone can agree on: they’re creepy as hell. But why? Especially when the AI ​​capable of recreating human faces seems to be much more complex. To find out, we decided to speak with a leading authority on the subject: AI.

ChatGPT by far the most famous AI language model today, which resonates greatly around the world with its ability to hold natural conversations, answer complex questions and create poems, extraordinary writing and even the most complex human discourse: game journalism. However, so far it is not capable of displaying images, music or videos, so a more impartial source in these areas.

ChatGPT on why AI struggles with fingers

I started by asking ChatGPT, “Why is AI so bad at showing human fingers?”

AI explains: “Rendering lifelike human fingers is a challenge for AI because they are highly articulated and have complex shapes and textures, adding, “capturing the fine The economics of how light interacts with skin, nails, and wrinkles requires advanced modeling and rendering techniques. “

While all of this is certainly true, it doesn’t leave out some key elements. First, the AI ​​is surprisingly adept at rendering the human face accurately, which is certainly a more complex structure than the hand, and—ah—the AI ​​can’t even comprehend it. numbers of the correct finger.

“A number of AI-powered tools and applications can generate realistic images of faces, including facial features such as eyes, nose, and mouth,” says ChatGPT. “The technology behind these apps is based on machine learning algorithms that have been trained on large datasets of human faces, allowing them to learn the patterns and details needed to create lifelike faces. . However, even with advanced algorithms and large data sets, rendering fingers remains a challenging task for AI.”

But why? Humans are extremely picky about faces, very quickly to immediately notice anything different from the standard of symmetry, but it can take a long time to notice something as obvious as a completely missing finger. It seems odd that the opposite is true for AI.

“Faces are relatively simpler to model because they have fewer moving parts and more uniform surface textures,” says ChatGPT. “The human brain is also very attuned to facial expressions and features, making the task of creating a convincing face a priority for AI researchers. As a result, AI has made significant advances in creating realistic faces, but rendering fingers remains a challenging task that requires further research and development.”

Read more: Oh no, ChatGPT has been cracked to be more reckless

So even though it might be as simple as a lack of prioritization by developers, it still strikes me as odd that it can’t even count to five. At this point, ChatGPT starts to exhibit some defensive abilities that tend to creep in when any AI is challenged. “While AI has made significant progress in creating realistic images, it is true that some AI models have difficulty generating the exact number of fingers in an image,” it began. First, again argue that the dataset may not be large enough. But then it continued, “However, it is worth noting that not all AI models have difficulty with finger rendering and some models were able to generate hand and finger images. very convincing hand. The development of advanced neural network architectures, as well as new techniques for training and data augmentation, could help solve some of the challenges in the future of realistic finger visualization. .”

So which one is succeeding? “One example is the Deep Hand Model, which is a generalized model that uses a 3D hand model to create realistic hand poses and shapes. Another example is the Self-Attention Adversarial Network (SAGAN), which has been used to generate a realistic image of the hand with a high level of detail and realism.”

ChatGPT about why fingers matter

But then, ChatGPT turned the conversation on its own into a more interesting realm. Because as much as we can question Why AI creates hands like impossible trident, it’s easy to stop asking why it matters. Luckily, ChatGPT was way ahead of me. “There are also AI models that specialize in generating specific types of hand images, such as hand gestures or sign language,” it said. “Example: A Sign Language Transformer (SLT) is an AI model that has been trained to translate sign language videos into natural language text, and it includes a hand posture estimator. can accurately predict the position and orientation of the fingers in the hand. “

According to the chat bot, SLT achieves the latter through a “convolutional neural network,” which then passes this information through “a series of transformer class,” which “encodes the space-time information of sign language gestures.” Finally, a “decoding network” generates the corresponding natural language text.

GPT adds, “It has the potential to improve communication and accessibility for the deaf and hard of hearing community, as well as other applications related to the interpretation of hand gestures.”

I asked if there were other practical uses to focus on better understanding human fingers and GPT provided me with the entire list.

1. Gesture recognition: Improved finger visibility could allow AI to recognize and interpret human gestures more accurately, which has many potential applications, including virtual reality, games, and more. play and robot.
2. Medical Applications: Advanced finger modeling and rendering techniques can be used in medical simulations to train surgeons and medical students in complex procedures, such as microsurgery.
3. Human-computer interaction: By accurately modeling the position and movement of fingers, AI can be used to develop more natural and intuitive interfaces for computers and devices other.
4. Security: Fingerprint recognition is a widely used biometric identification technology, and better finger visibility can improve its accuracy and reliability.
5. Accessibility: Improved finger visibility could help make assistive technologies more effective for people with disabilities, such as prosthetic hands or devices that allow easier manipulation of objects than.

With so many practical applications, especially so many of which benefit people with disabilities, I wonder if this makes focusing on human faces a bit weird.

“The fact that programmers focus so much on faces in AI research is not necessarily exceptional,” GPT replied, “because face recognition and generation are important research areas for many applications, such as such as security, entertainment and communication. The human face is also a complex and highly expressive part of the body, which makes it an exciting and challenging area of ​​research for AI.”

“That being said,” it continued before I could ask anything else, “is certainly valuable in exploring other areas of the body, such as the hands and fingers, while improving AI’s understanding and rendering of these body parts.“

Chatsonic on why humans are still the best

AI opponents chatsonic offers a much broader set of AI capabilities—although less likely to provide precise or coherent feedback (in a chat, I found it agree with me on the mental health benefits of having one’s own head amputated)—I think I should let it respond to the thread. I started by asking it to create a few images of human hands, which ChatGPT can’t do (though of course OpenAI, the team behind ChatGPT, also owns DALL-E). They are as creepy as we used to be.

Next, I also asked him why the AI ​​is so bad at this. “AI algorithms are still evolving, and they currently cannot accurately render a human hand in a realistic fashion,” says Chatsonic much more bluntly than ChatGPT usually offers. “AI algorithms are not designed to recognize the intricate details of human anatomy and movements, meaning they cannot accurately reproduce a human hand in a realistic way.” Though it added, with familiar confidence, “As AI algorithms continue to improve, it is likely that they will eventually reach a level of accuracy comparable to human hand-drawing.”

So I press again, why doesn’t it get the correct number of fingers? “AI algorithms are not yet advanced enough to accurately recognize the details that distinguish human hands from each other,” it said, and then with a humility you don’t often see from algorithm-based programs. According to this algorithm, “The anatomical nuances of the human hand require a level of precision and complexity that AI algorithms have not yet reached. This is why when it comes to creating lifelike digital replicas of human hands, humans are still the best choice for the job.”

I ended up asking both AIs if they thought perhaps their current helplessness was a benefit, as it provides an easy way for one to discern whether they are communicating with a person or an AI. The much friendlier Chatsonic was quick to agree. “Absolute!” it answers. “The inability of AI to accurately display human hands provides an invaluable way to distinguish between interactions with AI and interactions with real people. This helps create a safer online environment and allows users to know for sure who they are communicating with. Furthermore, it helps ensure that AI cannot take advantage of humans by posing as humans.”

Meanwhile, the endlessly tedious ChatGPT and its paranoid claim of constantly informing you of its lack of opinion, along with its gory-minded stance on “attacks”, have become should be much more popular.

“As an AI language model, I have no personal opinions or feelings. However, I can understand why some might see the difficulty the AI ​​has in accurately representing the human hand as a positive in that it provides a way to distinguish between the response of the human hand and the human hand. humans and AI,” it began, before continuing with four paragraphs on details I’d not been asked. Inside it all, however, it suggests a much more negative tone. “Ultimately, the goal of AI research is to develop systems that can perform tasks as well as or better than humans. While the limitations of AI in certain tasks may provide a way to distinguish between human response and today’s AI, it is likely that this distinction will become less obvious in the future. future as AI technology continues to advance.”