hi!!

H

hannahal

Filing Flight Plan
Joined
Nov 26, 2024
Messages
5
Display Name
Display name:
hannah
Hey there! Im Hannah Al, and im currently in college getting my basics. However, I want to hold different FAA certificates and learn about the mechanics of airplanes/jets, etc. For an assignment in class, I'm supposed to interview someone in the same career path as me, and I was wondering if anyone is willing or able to help me out.
thanks :)
 
i'm sorry i don't know a lot about the community-
 
This can be sort of a snarky group, but I'm sure someone will come along that will help you.
 
What career path exactly are you hoping to follow? I am neither a professional pilot or aircraft mechanic, so I can't help with the interview aspect. However, if you could clarify who you are hoping to interview, that could help others determine if they fit your desired profile. There are several members here who are either almost fully rated ATPs or flying airline pilots and at least a few mechanics.

Good luck on your journey - new pilots and/or A&Ps are a cause worthy of celebration. :)
 
SkyChaser said:
What career path exactly are you hoping to follow? I am neither a professional pilot or aircraft mechanic, so I can't help with the interview aspect. However, if you could clarify who you are hoping to interview, that could help others determine if they fit your desired profile. There are several members here who are either almost fully rated ATPs or flying airline pilots and at least a few mechanics.

Good luck on your journey - new pilots and/or A&Ps are a cause worthy of celebration. :)
Click to expand...
i would love if i could interview a flying airline pilot!!! it would actually make me so happy!! (also i need this grade really badly :,)
 
hannahal said:
i would love if i could interview a flying airline pilot!!! it would actually make me so happy!! (also i need this grade really badly :,)
Click to expand...
You might want to consider adding that to the title of your post or at least to the body of your first post, if not making a whole new thread with that as the title. You might be more likely to catch the right eye that way.
 
Agreed, this might help.
Consider change ‘Thread Title’

From: hi!!
To: College student, interview airline pilot for class assignment

… for a post in ‘Pilot Training’ forum.

New POA joiners may use ‘Start Conversation’ (for private chat/direct message), typically after 5 posts in public forum.
Hope forum tips are useful; good luck.
 
hannahal said:
i would love if i could interview a flying airline pilot!!! it would actually make me so happy!! (also i need this grade really badly :,)
Click to expand...

Airline pilots are required to hold an ATP certification, where ATP stands for Airline Transport Pilot. There are a few on this board, but this board is more focused on the GA community. GA stands for General Aviation and refers to hobby-pilots that fly for fun.

Good luck in your assignment.
 
Welcome! There are a few of us airline types around, although I'm not sure if any of us also worked on the mechanic side (I didn't, unfortunately). Feel free to start a conversation with me if you'd like, though. :)
 
Ashara Keliyn said:
It's just bunch of "if/then" statements.
Click to expand...
I don't think neural networks work like that. The programming of a LLM and all its possible inputs and outputs using branching (like you do with IF/THEN statements) would be incredibly cumbersome and limiting at the same time.
 
Baked Potato said:
I don't think neural networks work like that. The programming of a LLM and all its possible inputs and outputs using branching (like you do with IF/THEN statements) would be incredibly cumbersome and limiting at the same time.
Click to expand...
Well, fundamentally down at the lower levels, compare and branch is a necessary operation to implement a LLM.
 
Salty said:
Well, fundamentally down at the lower levels, compare and branch is a necessary operation to implement a LLM.
Click to expand...
The lowest level I'm familiar with is a node, each of which is weighted as to how much it influences nodes at the next layer. And the relationship is not always linear. See the attached image. How would you define the activation of a boundary for the second image in the picture, using IF/THEN statements? And keep in mind that for a LLM that is designed to respond to a wide variety of inputs (the entirety of the English language, for example, or input of any image) then there can be a seemingly-infinite number of these.

Furthermore the entire basis of deep learning is that it does not require human intervention at that level - the weighting of the nodes is adjusted by the model itself as it verifies the outputs against some measure of accuracy. IF/THEN statements at that granular of a level would obviate the model's ability to adjust itself as needed to fit different scenarios or to improve output. Take domain-specific training, for example. If you're training an LLM for your own purposes (like a chat bot to help your employees navigate your own company documentation, for example), would you want to go in and adjust the parameters for a million IF/THEN statements that define the activation function for each node?
 

Attachments

  • Screenshot_20241130-122653.png
    Screenshot_20241130-122653.png
    717.1 KB · Views: 8
Last edited:
Baked Potato said:
The lowest level I'm familiar with is a node, each of which is weighted as to how much it influences nodes at the next layer. And the relationship is not always linear. See the attached image. How would you define the activation of a boundary for the second image in the picture, using IF/THEN statements? And keep in mind that for a LLM that is designed to respond to a wide variety of inputs (the entirety of the English language, for example, or input of any image) then there can be a seemingly-infinite number of these.

Furthermore the entire basis of deep learning is that it does not require human intervention at that level - the weighting of the nodes is adjusted by the model itself as it verifies the outputs against some measure of accuracy. IF/THEN statements at that granular of a level would obviate the model's ability to adjust itself as needed to fit different scenarios or to improve output. Take domain-specific training, for example. If you're training an LLM for your own purposes (like a chat bot to help your employees navigate your own company documentation, for example), would you want to go in and adjust the parameters for a million IF/THEN statements that define the activation function for each node?
Click to expand...
You completely missed my point. It wasn’t what you responded to.
 
Salty said:
You completely missed my point. It wasn’t what you responded to.
Click to expand...
Can you elaborate then? What else do you mean by "fundamentally down at the lower levels"?

Circling back to my original response to "AI is just a series of IF/THEN statements", which can be true for non-generative AI (like a chess bot or video game NPC) but is certainly not used for construction of LLMs. Here is an example of how it can be done using Python. It uses mathematical functions that are higher than my pay grade. Even if you are somehow able to logically define the activation of each node using IF/THEN statements, explaining away the entire process like the original post I replied to is like answering the question of "how do airplanes fly" by saying "it's just a bunch of pistons going up and down".
 
Last edited:
Baked Potato said:
Can you elaborate then? What else do you mean by "fundamentally down at the lower levels"?

Circling back to my original response to "AI is just a series of IF/THEN statements", which can be true for non-generative AI (like a chess bot or video game NPC) but is certainly not used for construction of LLMs. Here is an example of how it can be done using Python. It uses mathematical functions that are higher than my pay grade. Even if you are somehow able to logically define the activation of each node using IF/THEN statements, explaining away the entire process like the original post I replied to is like answering the question of "how do airplanes fly" by saying "it's just a bunch of pistons going up and down".
Click to expand...
All that fancy logic compiles down to if then statements in the end. No, it’s not coded that way, but inevitably that’s what it is.
 
Salty said:
All that fancy logic compiles down to if then statements in the end. No, it’s not coded that way, but inevitably that’s what it is.
Click to expand...
IF/THEN implies human-readable programming (i.e. a person creating a branching statement in a programming language). Compiling is the conversion of that to machine language, where IF/THEN does not explicitly exist, to my knowledge.
 
Baked Potato said:
IF/THEN implies human-readable programming (i.e. a person creating a branching statement in a programming language). Compiling is the conversion of that to machine language, where IF/THEN does not explicitly exist, to my knowledge.
Click to expand...
In machine language it’s compare and branch as I said in my first post. But it’s still an if then.
 
Salty said:
In machine language it’s compare and branch as I said in my first post. But it’s still an if then.
Click to expand...
Is it? LLMs are based on probability and not definitive Boolean logic. Based on the complexity and sheer number of probabilistic decisions we all know that LLMs are non-deterministic. An interesting question is whether or not one can even create a non-deterministic system using IF/THEN statements without shoehorning something artificial like a random number generator into each layer.
 
Baked Potato said:
Is it? LLMs are based on probability and not definitive Boolean logic. Based on the complexity and sheer number of probabilistic decisions we all know that LLMs are non-deterministic. An interesting question is whether or not one can even create a non-deterministic system using IF/THEN statements without shoehorning something artificial like a random number generator into each layer.
Click to expand...
Even LLMs run, in the end, on digital hardware that uses Boolean logic at its core.
 
DaleB said:
Even LLMs run, in the end, on digital hardware that uses Boolean logic at its core.
Click to expand...
Yes, but that's reductive and has nothing to do with how AI actually works. Nobody talks about IF/THEN statements at a transistor level. At that point it's TRUE or FALSE and we're talking about individual bits.
 
William Pete Hodges said:
Sorry to sound stupid here but what is an LLM? It sounds like a brand of cigarettes to me.
Click to expand...
Large Language Model. It's the system that's used in generative AI applications like Chat GPT and Copilot. It is able to parse natural language and respond in kind. It differs from the types of AI that have been used for decades that are simply a series of branching, hard-coded decision points.
 
You must log in or register to reply here.
Back
Top