How can I make software like Jarvis

Can JARVIS be real?

After you've worked on AI, the correct answer is

Although I agree with other replies that you are JARVIS to a certain degree and have it follow simple preprogrammed commands, this is VERY different from what JARVIS actually does.

Let's start from the beginning. In the current top answer, the functions of JARVIS are listed as such

  • Jarvis can speak
  • He can wake you up
  • Jarvis watches over the house
  • It can even help with migraines

All of this is easy (not really, but easier than what I'll describe later) to program into an AI agent.


Speech recognition and synthesizers are becoming more common (see Siri and Google Voice). They hear audio, filter noise and convert speech into text. Then they use text analysis (like MALT) to extract dependencies in the text. You compare them to hundreds of ready-made commands and customize them accordingly. They perform any task selected by the user ( preprogrammed Tasks) and respond with voice-text software (the oldest of these technologies! Today you can listen to Google Translate).


Ok, that's stupidly easy. You choose any calendar (Microsoft or Google or whatever) or alarm system (even from your phone!) And it can do it. It's a standard technology in every smartphone these days.


This is a more complex technology that is not as readily available. Smart houses become something (Bosch invests a lot in this area, for example) and have the skills you described. They monitor who is in the house, temperature and light, locks, even food in the refrigerator. These, too, are mostly preprogrammed tasks that the system ultimately learns to learn using machine learning techniques at appropriate times.


There are also systems that track health and the like. For example, some smartwatches track your heart rate and it's not a big leap to assume that they could be configured to set off an alarm if the heart rate suddenly changes.

Apart from that, JARVIS does something completely different.

JARVIS can argue . Real, current, reasoning. He can have a conversation with you, request new information, assign conclusions, filter, discard and calculate. I can't stress how revolutionary this is. Of all the things Tony Stark invented in his comics, the most impressive is an Artificial General Intelligence (AGI) like JARVIS.

Today's research only captures how a machine can learn to think. It is something that has always been programmed by humans, not something that a machine has ever learned from a simple analysis of data, as we humans do. Machines can learn information from data, e.g. B. Show pictures of cats and dogs, and after a while the machine can distinguish between them. However, if you show them a bunch of balls of different sizes in a box and ask "which ball is furthest from the largest ball," it is immensely difficult to learn.

Google DeepMind (the leading research group on neural AI) published two articles in the last few months in which the machine learns to think. Check them out here and here. If you read through it, you will see how groundbreaking the research is and how similar it is not. An AGI like JARVIS is astonishingly more advanced than these efforts, and it will take years until we can achieve that level of autonomy, efficiency, and general intelligence.


They say that JARVIS is able to argue. But we cannot know for sure. JARVIS may simply have been programmed to check Tony's behavior. This then means that JARVIS reminds Tony of something that Tony forgot. E.g. when Tony programs Jarvis to remind him 5 minutes before the oven timer runs out; and we only see JARVIS telling Tony that "his meal is almost ready"; We (as viewers) might mistake JARVIS 'preprogrammed message for JARVIS actually taking care of Tony's food quality. Provided that JARVIS is sufficiently developed, most of its behavior could be preprogrammed, seems but to be real.


If Tony has had problems with the power of his suit burning out in the past, he may have updated JARVIS to look for any signs of this. We assume, that JARVIS decided to do this themselves (see to Tony's safe return). But it is possible for Tony to update JARVIS if he makes another mistake so that JARVIS makes the observation and Tony can not worry. If you assume Tony makes enough mistakes and updates JARVIS; this removes the notion that JARVIS is doing things because it has argued that it is useful for him to do so.


@Flater Sure, you can assume Tony hardcoded everything into JARVIS, but since this is a comic book and it would take an astronomical long time, this is not likely. In Marvel's universe, I find it more plausible that JARVIS is an AGI since Stark is a genius. That's why it can save Stark without asking why it can fight Ultron, escape to the internet to rebuild itself (as if that was actually a thing), and come back to be assimilated by Vision why it can analyze crime scenes and infinity stones etc. etc. But in fact there is no official "Word of God" that JARVIS is an AGI.