The course appears to be good, but I am confused on what AI is…understanding the machine learning, but what makes it work? Is it programming language or a software I can buy? What is it physically? Thanks!
1 Like
I believe this will be explained as you continue past Week 1.
2 Likes
Hi I’m so glad I’m not alone,same my side
1 Like
I understand but don’t know how to begin to start putting it together
1 Like
Morning all, can anybody perhaps tell me how should week 1 assignment look like and give me a idea how to start off with mine please if possible anyone
The best place to post questions about a specific course is in the Forum area for that course.
- If it’s a Short Course, use the “Short Course Q&A” area.
- If it’s a regular course (through Coursera), use the “Course Q&A” area.
In either case, select your course, then the “+ New Topic” button, and add a Week # tag.
1 Like
AI is a Methodology. A set of Tools and Techniques that primarily uses the principles of Machine Learning - Supervised/Unsupervised to analyse Data and draw meaningful results from it. It “uses” different Technologies for this purpose. For rest, even I am almost a beginner in this journey and things would get clearer as we proceed.
2 Likes
AI, or Artificial Intelligence, is about making machines “smart” so they can solve problems or make decisions like humans do. Machine Learning (ML) is a type of AI where computers learn from data instead of being directly programmed for every task.
What makes AI work? It’s a combination of:
- Data: Lots of examples to learn from.
- Algorithms: Step-by-step instructions for finding patterns.
- Programming: AI is often built using coding languages like Python.
- Software: There are tools (like TensorFlow or Scikit-learn) that help create AI models.
AI is not something you can touch—it’s code that runs on powerful computers. Think of it as a recipe (the code) and ingredients (the data) that together create something amazing!
1 Like
Which course it is? Thanks!
I believe this thread is in regards to:
1 Like
My emphasis added above
I know this appears often in discussions like this, but I don’t think it adds value and try to avoid it. I don’t believe anyone completely understands how humans solve problems. To piggyback on your food metaphor, I’m certainly not running gradient descent when I contemplate whether or not salt needs to be added to my soup.
AI is not based on magic or intuition- it’s based on matrix subtraction and computing partial derivatives. Pretty sure that’s not what is happening in my brain. It’s for sure not happening in my 2 year old grandson’s brain.
1 Like
Disagree.
That is an inspiring answer, @TMosh .
Biologists don’t completely understand the ‘thought’ process but their partial understanding is applied in AI to create Artificial Neural Networks (‘artificial’ being the key term) that approximate (not exactly mimic) how our thought processes work and enable us to solve problems. That is the main concept behind AI.
You don’t run gradient descent (GD) but GD approximates how a human would solve a problem. For clarity, how do you balance on a slope? By leaning opposite to gravity/slope. If you overcorrect you fall down, if you undercorrect you fall down, so you try to do the opposite in correct measures of what is detrimental to your staying on your feet. An AI agent will mimic this as GD. So yes, we don’t do GD but coders and mathematicians use the algorithm to achieve something similar to what a human would do. This is why it is called ‘Artificial Intelligence’ and not ‘Human Intelligence’.
Welcome, @sofi_z .
1 Like
Achieving similar outcome doesn’t necessarily imply nor require similar process to arrive at said outcome. In my personal professional experience using language like solve problems or make decisions like humans do established false and unrealistic expectations about project cost and complexity from customers. In my opinion including it in a description of AI doesn’t add clarity but it does add ambiguity and this is why I don’t recommend it.
Cheers
1 Like
@ai_curious , customers in industry have unrealistic expectations because they don’t understand that technical terms in the AI field may mean something different in their own fields & of course service providers do false advertising.
Yes, AI is about ‘machines’ solving problems and making decisions like humans. Classification is an unsupervised technique where your code/algorithm will put like objects in the same group, separate from other groups. An example of make decisions like humans is having a reinforcement learning algorithm that can learn to play chess and eventually win against world champions. In all these cases the AI follows algorithms that are based on what a human would do to solve the problem. Remember that English words and many words in other languages may have slightly different meanings. No one word has exactly one meaning in language usually. In the realm of computer science, statistics and mathematics it is agreed by the majority of experts and students (textbooks, tutorials, etc) that AI solves problems and makes decisions like humans do. I am sure that most books on AI will point out that ANNs are approximations of the human biology which we still don’t fully comprehend. That disclaimer is there. Almost every AI student knows this or is taught this as an introduction to the field. When you are in a specialised field you use the language of that field and the meaning of that language is relevant for that field-it may be difficult to extend it to some other field. English language definition of ‘vacuum’ is ‘empty space’. Quantum Mechanics definition of ‘vacuum’ says ‘not empty space, but filled with virtual particles that are in a constant state of existing and not existing’. You can’t reconcile those two definitions but it is clear that the one word ‘vacuum’ means something different to different audiences. Use the accepted terms where they belong.
Do you have a better description of AI that you recommend to @BarbaraK and @ptrikha10 ? It will really be helpful to provide that.
Hi @Rorisang
Just like you shared a definition of vaccum from English dictionary to quantum mechanics, artificial intelligence also is defined by different institutions or certified people who have worked on AI, and almost every definition if you check none tells AI tries to find a solution or compute algorithm just like human do, rather it states AI is a capability of algorithm or computation configuration trying to solve an issue based on combination of mathematical, statistical as well as intuitive capability trying to mimic human intelligence yet even Prof. Ng mentions no matter what how a human brain works nobody can completely understand on why a kid choose one day chocolate over pizza. So some part of the intelligence cannot be completely stolen from human to machine but AI trying to understand or do a computational probability of how many times a kid might choose chocolate over pizza is something of creative idea where AI can try to comprehend but doesn’t assures it will choose chocolate today and tomorrow not, and that’s why when one creates a model the accuracy is never about 100% but 99% as even human are not a perfect being nor machine.
Regards
DP
Can you please look up the definition/concept of reinforcement learning and see if it does not oppose what you say in the quote?
https://aws.amazon.com/what-is/reinforcement-learning/
From the same site:
The learning process of reinforcement learning (RL) algorithms is similar to animal and human reinforcement learning in the field of behavioral psychology. For instance, a child may discover that they receive parental praise when they help a sibling or clean but receive negative reactions when they throw toys or yell. Soon, the child learns which combination of activities results in the end reward.
An RL algorithm mimics a similar learning process. It tries different activities to learn the associated negative and positive values to achieve the end reward outcome.
Reinforcement learning algorithm is one of the techniques used to improvise the accuracy by using RLHF but it no where tells its adding this algorithm by its own reasoning thinking but human adds a data in ways so the algorithm understand how human might work, that doesnt confirm human might work the same.
Reinforcement learning is only used in cases where complete reasoning can be provided but sometimes even you must be know what a human does can be out any reason or logic, can AI catch this ??
So AI doesn’t work like humans do!!! its brain behind AI algorithm which is combination of logical as well as practical and also not to forget human Intuition.