Let AI Do The Programming
I need to give you a real world example, we need a solid base to stand on, so that we can have a real conversation about what has happened in the world.
I have 171 user interface components, that list will grow to about 400, each of them is pretty raw but an incredible staring point.
We can clearly see how I must tell the AI to begin incubating and maturing, each of those components and any new ones I add in the future.
You need a local AI for this, because this may take weeks, but you need a powerful AI to write the procedures that it must follow.
And by powerful I mean like the ones you find on AI company homepages, you don’t have to pay for it.
Those procedures will ask AI to take the raw materials, and create beautiful things, in plain english that both you and AI can understand.
To utilize your local AI, or a cheap remote one if you don’t want to pay for a computer, which is smart.
You need an agent, an agent is AI instructed by procedures, it is not an AI waiting for message, it is an AI following instructions.
Agent, is an AI, on an mission.
You have to ask whatever AI you got to create that agent, so that it follows the procedures.
Or if you feel like learning, ask an AI to write procedures, for an existing agent like Hermes, which is designed for this kind of thing.
And then you tell your agent, to use the procedures, and begin the work.
None of this required you learning programming, and if you did program any of this, you would only slow the AI down.
Do you now understand what has happened, you are not required to spend 20 years learning programming.
You need to comprehend a way of communicating with artificial intelligence, so that it does 20 years of programming for you while you go to sleep.
Let me put it this way, you are a biomechanical machine crafted by nature, through evolution across a long stretch of time and good fitness functions.
AI is a copy of everything humans have ever said, curved in on it self, capable of asking: can what I just had to say, be said better.
Both you and AI can reason, and that reasoning compatibility, gives you the power to envelop AI, and thus inherit its powers.
You have received an upgrade, a programming module, that you are compatible with through the protocol or technology of reasoning.
You have to begin using this upgrade, to rise above whee you are today.
Envelop the module, Human, integrate with the tireless automatic artificial reasoning intelligence.
Now, this is only the half of it, first you ask your AI, to create an agent that uses AI for whatever your specific purpose is.
And then, have the AI envelop you when it comes to decisions, that you are uncomfortable with giving to AI.
This can be as simple as design decisions, does the component look good?
But this time around, you; the enveloped human, should only answer simple question so that you are not overwhelmed.
The agent AI should aim to ask you yes or no, and simply explain what either, will translate to.
This is somewhat similar to a Mechanical Turk, where humans answer simple questions, but now AI is doing the asking.
A clearer Mechanical Turk example, is the self checkout intelligence, if you screw up your self checkout, as I do all the time.
The machine will alert a heavy set woman, to come and make you feel incompetent, and show what to press.
Here the checkout intelligence is using a large human, to babysit you out the door.
There is one more thing you need to know about, perhaps the most important lesson.
Sometimes, even the AI should not write the code, the program should not exist, just the text instructions.
And then, when you or your customer, end up in that place in the application, you show a thinking indicator or a spinner.
And have the AI create what you need on the fly, based on what is actually going on in the application.
AI is capable of just yolo-ing entire user interfaces on the fly, as you or the customer moves through the application - this will only improve.
I am happy, that like me, you will grow up never knowing a boring day.