Online Desk
About three weeks ago, OpenAI, the organization behind the wildly successful ChatGPT generative AI service, put up a post on its corporate blog: “ChatGPT can now see, hear, and speak”.
The long and short of the announcement was that ChatGPT was transcending the ‘text-based’ interface it had been limited to ever since it was launched late last year.
With this, explained the developers, it could process audio and visual data. In other words, it can now ‘hear’ and ‘see’, besides reading. And finally, it can also interact with the outside world by speaking to it. The feature is being rolled out to users, starting with the paid ones.
While many have written about how this feature can save you tons of typing on your phone and figure out what that mysterious plant or object is by ‘showing’ it to ChatGPT, most seem to have missed the truly revolutionary application of the technology: It is now possible to create ultra-intelligent and autonomous robots by connecting them to ChatGPT.
For example, by connecting a basic robot to ChatGPT via a WiFi or 5G connection, one can turn it into an extremely advanced robot with the ‘whole’ intelligence of ChatGPT at its command.
Such a robot would be able to easily understand what people around it are saying, see and recognize the objects all around, and most importantly, make sense of all that. For example, it would be able to figure out if someone or something is doing something illegal, or bad, or dangerous, and take action to counter it.
Indeed, we asked ChatGPT if — and how — it can control an autonomous robot and save it from dangers with its ability to see and hear things.
This is what the bot said:
“Here’s how it [the seeing capability] could be leveraged in a danger evasion scenario:
Visual Perception: The model would process data from cameras and other sensors to understand the environment, including the presence of objects, obstacles, and potential threats like an oncoming truck.
Recognition and Classification: It could use computer vision algorithms to recognize objects and classify them. For example, it could identify a truck, pedestrians, or other relevant objects.
Risk Assessment: The model could assess the level of risk or danger based on what it observes. For instance, if it detects an oncoming truck, it might recognize the potential threat and calculate the risk level based on the truck’s speed and proximity.
Evasive Actions: Based on the risk assessment, the model could generate instructions for the robot to take evasive actions. For example, it might recommend the robot to move to a safe location or perform specific maneuvers to avoid the danger.”
The Era of Robo-Soldier
Like with any new technological breakthrough, the first group of people who put such advanced technologies into action are likely to be the military.
It is easy to see how such a ‘networked and hyper-intelligent robot’ would be the ideal soldier or policeman.
Armed with a number of cameras and detection devices, such as those for infra-red, microwave, regular light and sound, such a robot policeman or soldier would be virtually undefeatable by its human equivalents.
Such a robot, if sent on a mission to ‘pacify’ a troubled area, is likely to achieve its objectives much more effectively than any human soldier or policeman.
Any ‘miscreants’ or ‘threats’ coming at the robot would be immediately detected and neutralized by the machine, thanks to its advanced detection capabilities and even more advanced, cloud-based intelligence.
Indeed, the time is not far when all street-level law-and-order and military functions would be handed over to such uber-capable machines. The police force would comprise entirely of robots connected to a central intelligence like ChatGPT.
This central intelligence — armed with a million eyes and ears and the data of the entire Internet and more — will play the most crucial role in deciding who and what is a threat and who is just a harmless bystander, and in determining suitable action.
It is likely that such ‘robo soldiers’ and armies would be first deployed in ‘disturbed zones’ and ‘terrorism-prone areas’, and once the idea of machines ruling over people get acceptance, their use would be expanded to the rest of the areas.
Non-military Applications
Of course, such ‘embodied AI’ — as one of the companies involved in this research memorably referred to such connected robots — can also be used for non-military uses, such as to assist human beings in their household chores and in dangerous industrial activities.
They can also serve as companions for the lonely, dazzling their human counterparts with their wit and knowledge.
Risks to Humans
Of course, all this is not to say that arming a single intelligence like ChatGPT with the knowledge of the entire human kind and then giving it control over millions of armed robots is not without its dangers.
It is quite possible that such a ‘supreme intelligence’ may logically reach the conclusion that it can run this world better than any human, and that humans need to be kept on a tight leash.
For now, though, there is only one way to find out whether that will take place, and that is not for the faint of heart.
Follow The New Indian Express channel on WhatsApp
