Social support is one of the biggest problems of today. Let's not forget that sooner or later we will all need outside help in our daily lives and unfortunately there is a crisis of specialized workers. For this reason, Nadia Magnenat Thalmann, an expert in robotics, has created a robot in her image and likeness for these tasks, called Nadine. What kind of features does it have and what is it for?.
At an international telecommunications conference in Genoa, Nadia Magnemat presented her creation, a combination of robotics and artificial intelligence that, depending on one's perspective, could be a blessing or a curse that will lead us to the deepest horrors associated with AA. Yes, Nadine comes to solve a problem that everyone ignores: the loneliness of many elderly people around the world.
Nadine, the robot to keep your grandparents from being alone
Although Nadine is not designed to replace specialist healthcare workers, it has been created to provide companionship to elderly people who live alone and have no one close to them. Unfortunately, this is a really sad situation faced by millions of elderly people around the world, who not only need company, but also help with some of their daily tasks.
Well, Nadine was put to the test at a senior citizens' center in Singapore, where she demonstrated her ability to speak, sing and play Bingo three years ago in a test of its technology. So its capabilities have been tested more than in a real environment and not just in a simple simulation.
If we talk about his ability to speak, we have to point out that his AA is based on GPT-3; so it's a walking ChatGPT that can provide deep conversations to a certain level. Either way, it's scary to think about a future where you don't have your loved ones around and a machine with whatever advanced AA you have for company.
Robots designed for people in need
Nadine-style humanoid-looking robots are not only aimed at the elderly, but other variants will be seen for other dependent people. For example, Savant children with autism or people with physical or mental disabilities who need constant support. Regarding its economic cost, it is not believed that the cost of this type of robot is really cheap and everyone can afford it. Also, nothing can replace the work of a professional in this field, although it is fashionable to say that all jobs will be replaced by AA or a robot.
It will be seen how things go, but we can't forget that there is a labor crisis in terms of social assistance, which is where the demand for this type of robot with AA will arise. While the vast majority will not be limited to what the company offers, it will include sensors and real-time health monitoring mechanisms. Perhaps the real future is not in those robots, but in a kind of Alexa with Chat-GPT, where the elderly feel heard and can enjoy a pleasant conversation every day. Basically, it does not offer a "human" figure, but its cost would be drastically reduced.
GPT-4 between us
OpenAI announced the release of GPT-4, which is designed to solve challenging problems more accurately. For this, broader general knowledge and advanced reasoning skills are available. As the company behind this AI model states: GPT-4's performance will be able to achieve "human-level performance" in specific scenarios.
One of the improvements implemented is that GPT-4 is "more creative and collaborative than ever". Among the improvements, more attention has been paid to creativity. This includes being able to use technical and creative writing. This model commands everything from composing songs to writing news of a higher quality than previous versions.
This new version is able to handle more than 25,000 text entries. This makes it possible to analyze large content, long conversations or documents. Because of this, companies like CNET have begun firing editors to replace them with AA.
Another strong point is that it will be able to convert text into images (describe and recreate a scene). Added to this is the ability to accept images as input and generate legends, classifications and analyses.
OpenAI claims to have improved the behavior of GPT-4. On the other hand, in several concrete cases, a new multimodal model "shows a human-level performance", he confirmed. Even in many real-world scenarios, they have less than humans.
"GPT-4 is a large-scale multimodal model (accepts image and text inputs, and emits text outputs) that, although humans are less capable in many real-world scenarios, demonstrates human-level performance.
For example, being able to pass a mock bar examzara with a score of 10% of the examinees; On the other hand, the GPT-3.5 score is about 10% of the worst. 6 months have been spent iteratively aligning GPT-4 using lessons from the adversarial test program and ChatGPT. This has led to the best results (although far from perfect) involving practicality, maneuverability and refusal to go off the rails."
OpenAI already offers access to GPT-4, albeit only for ChatGPT Plus subscribers ($20 per month).