Asking The Right Questions

We’ve all utilized artificial intelligence in some form since January 2023, whether asking ChatGPT to complete history homework or Claude for recipe ideas. However, many people seem to take it for granted as they do not appreciate the fundamentals of artificial intelligence. They might not even know what AI is; they use it as a tool to get things done. As education continues to be disrupted, teachers are handing Generation Z students multiple-choice tests and packets. In class, they are told not to take advantage of this 4th industrial revolution for essay examples and ideations. Instead of using Artificial Intelligence to complete useless homework and assignments, students should focus on absorbing the actual capacity of what AI can bring forth and utilize the tool alongside future passions or intended majors, for example.

Let’s use the space and community of fashion as an example. A big crisis in this field is the state of fast fashion; 85% of this overproduction ends up incinerated or landfilled annually. This linear disposal model depletes resources and pollutes our ecosystems.

Suppose I wanted to run a night market, selling pieces from recycled materials and thrift clothes. In that case, I can ask AI questions such as, “How can I make interactive stations for the guests?” or “How could I make the event more visually appealing?” to boost creativity and improve the quality of my process and the outcome of this event. Questions like these can help aid us in our creative work, just as long as we ensure these tools do not take over our work.

If Gen Z educated themselves more on chatbots, we could substantially grow our digital world and take advantage of Artificial Intelligence. We could even push beyond the current 4th industrial revolution. If younger generations do not grasp how AI functions, society faces unchecked technological development that deprives humanity of capturing AI’s immense problem-solving potential as we approach the future.

What is AI?

In its simplest form, Artificial Intelligence (AI) refers to intelligent behavior exhibited by software and machines, contrasting with natural intelligence demonstrated by humans and animals. It uses math, computer-based algorithms, probability, and machine or deep learning to mimic and output human recognition, decisions, and capabilities.

How Does It Work?

It learns these human cognitive functions by analyzing patterns, large data sets, or specifically told information. With this information and past experiences and mistakes, AI can discover the best parameters for a model to predict future outcomes accurately. Additionally, AI attempts to work with various outputs, whether voicing over someone or illustrating an image with parameters. Many notable examples of AI include Siri, NPCs in video games, and self-driving cars. Various categories related to the topic include robotics, natural language processing (NLP), expert systems, and computer vision.

Artificial Intelligence often relies on human supervision and labeled data to improve AI systems. For example, consider the case of an AI-Enhanced Waste Recycling & Sorting System designed to optimize waste sorting within recycling facilities. Human input and training are essential in these systems. Humans perform tasks like labeling data, refining AI algorithms, and continuously monitoring the sorting process. This human involvement is necessary to correct errors and ensure accurate sorting, preventing situations where materials like Styrofoam could be wrongly classified as recyclable. This crucial process, which allows AI to learn and improve, is known as machine learning, a subset of the broader field of Artificial Intelligence.

In contrast, deep learning, a subset of machine learning, excels in recognizing and adapting to patterns in games and complex applications like image recognition and natural language processing. For example, in a soccer sports video, using the same strategy repeatedly can make it easier for the opposing team to counter your moves, potentially leading to conceding a goal. In such cases, the game might employ AI techniques to keep the game challenging and dynamic. Deep learning is used to recognize and adapt to patterns, enhancing this gaming experience.

There is no doubt that AI is advancing; its impact rate is so fast it's nothing nobody has ever seen since the Industrial Revolution. It's unprecedented and scary, given that AI must absorb mass amounts of data to notice patterns or specific data to provide a correct output. The revenue gained from generative AI continues to grow exponentially, too. 

"The global artificial intelligence (AI) market size was valued at USD 454.12 billion in 2022 and is expected to hit around USD 2,575.16 billion by 2032, progressing with a compound annual growth rate (CAGR) of 19% from 2023 to 2032. The North America artificial intelligence market was valued at USD 167.30 billion in 2022."

Many other AI tools have surfaced since OpenAI introduced ChatGPT 3.5 to the public. The capabilities continue to grow as well, astonishing the masses.

Google has just recently demoed its brand new AI model, Gemini. The company constructed Gemini to be 'multi-modal,' with the capacity to take in data not solely in text form but also in the form of images, audio, and video.

Dora AI is a new website design program that uses artificial intelligence to make websites. It can create a website just from one line of text. The creative maker is being tested and can seemingly make intuitive and creative dynamic 3D designs from straightforward lines like "make a website for SpaceX starship"

According to an article by Forbes, automation and AI will eliminate 83 million jobs globally by 2025. Forbes calls it a "new era of turbulence" since many jobs will look different or disappear. Jobs like data entry and admin support will decline a lot. Luckily, the World Economic Forum claims that 97 million new jobs will be created, too, mostly in tech, healthcare, and green energy. Still, communities hit hardest by job loss will need support from the government and businesses, so it will be tough for many workers to keep up with all the changes happening so fast. Creativity, flexibility, and critical thinking will be even more valuable. Data, AI, and content creation jobs will grow; therefore, for the next five years, people will have to constantly 'feed' themselves with these new or similar abilities.

If Gen Z wants to prosper and excel in the current day with rapid improvements in AI, they need to make sure they aren't easily replaceable by AI, and not more than just jobs; people as a whole.

Everyone has their voice, their agency, their style. If AI copies a person's characteristics or traits, they might replace that person entirely as a robot in their jobs, daily routines, and much more. Additionally, Gen Z needs to get creative about their passionate pathways, as many can be expendable by AI soon.

Millions of people will lose their jobs because AI will replace them. Suppose Gen Z cannot hold onto their identity and agency while thriving within their pathways. In that case, they will all be as unlucky as anybody else until, sooner or later, AI will enslave us all. We must ensure this does not happen to our generation; otherwise, we are all doomed.

If you don't already know, math is the magic behind Artificial Intelligence. But what kind of math is prominent in AI, and how does it work to give you your desired answers?

Linear algebra is the first significant mathematical foundation underpinning the design of neural networks in artificial intelligence. It involves two pivotal structures: vectors and matrices. Vectors are directed line segments represented by an arrow with two attributes - direction and magnitude. Some examples of common vectors include velocity, force, momentum, weight, and electromagnetic fields. Vectors provide a mathematical representation of data, enabling AI systems to discern meaningful connections, perform complex calculations, and bring structure and meaning to information. A matrix is a two-dimensional rectangular array of numbers arranged into rows and columns. The dimensions of a matrix indicate its size - the number of rows followed by the number of columns. Matrices allow AI algorithms to compute linear transformations and other advanced mathematical operations efficiently. This mathematical machinery equips AI models with the capacity to analyze data and uncover functional patterns.

In machine learning, we use multivariate calculus to optimize or "train" models - tweaking them iteratively to improve performance. One essential technique is gradient descent. By calculating the gradient, which quantifies the rate of change of the model's error as a function of its parameters, we can determine how rapidly adjusting each parameter would increase or decrease the error. You can aggregate these parameter gradients into a Jacobian vector. Leveraging this vector enables the simultaneous updating of all the model parameters in an error-reducing direction. Deep neural networks contain multiple layers and parameters, so propagating gradients backward from the final output layer to the initial input layers (backpropagation) efficiently computes the influence of each parameter on the last error. The Jacobian vector then allows updating all weights in the network simultaneously to reduce mistakes.

Multivariate calculus is essential for AI to operate on high-dimensional data and models. It furnishes critical techniques like partial derivatives, gradients, and Jacobian matrices to optimize functions with multiple variables. This mathematical machinery empowers methods like neural networks, robotics, and computer vision to refine models and boost performance. In summary, multivariate calculus provides the core mathematics for AI systems to tackle complex real-world problems.

Artificial intelligence relies heavily on probability and statistics to interpret noisy, real-world data. Probability enables AI systems to determine whether observed patterns represent meaningful signals or random noise. Statistics provides AI tools to summarize data, test hypotheses, and assess machine learning models. Fundamental concepts like measures of central tendency, variance, and probability distributions give AI the mathematical vocabulary to characterize patterns. Bayesian inference allows AI systems to update their beliefs as new evidence becomes available.

Probability and statistics supply the mathematical framework for AI to grapple with uncertainty and generate useful predictions - critical capabilities that distinguish AI from rigid, deterministic programming. By incorporating probabilistic and statistical reasoning, AI systems can better emulate human learning and decision-making.

Now that we've covered the math that has the most priority, I provided both ChatGPT and Claude Ai with a generic question that anybody could potentially ask:

"Can you recommend a couple icebreaker activities for a team-building day at work? We have remote employees, and want to make them feel included."

I then asked how all the math we mentioned in AI works together so that these chatbots give precise answers and the process behind it. I also asked them to summarize the significant differences and similarities with the other chatbots. I wanted to see how the pieces of math worked together to produce an outcome for such a question.

Here are their responses:

ChatGPT

Claude

ChatGPT

Claude

In tackling the challenge of recommending icebreaker activities for a team-building day with remote employees, both ChatGPT and Claude.ai share vital similarities and differences. Both systems use natural language processing to break down the question, identify crucial elements, and consult a knowledge base or training data for relevant examples.

Both systems employ probabilistic models to evaluate and rank potential activities based on suitability and generate a natural language response. However, ChatGPT delves into specific machine-learning algorithms, emphasizing the importance of a massive dataset for learning patterns. In contrast, Claude.ai provides a simplified explanation without naming specific algorithms, focusing more on mathematical representations and operations like vectors and linear algebra.

ChatGPT emphasizes the importance of learning from an extensive dataset, while Claude.ai focuses more on the mathematical aspects. ChatGPT provides more details on the evaluation process, considering multiple criteria, while Claude.ai describes a general probability-matching process.

While private high schools such as Harvard West-Lake have rapidly welcomed AI by providing related courses and encouraging its use, public high schools have been much slower to take on this tech; some teachers even go as far as to say not to use it at all. My 11th grade AP Language & Composition Teacher went as far as not to allow us to utilize the internet while writing essays for examples for our writing; it’s ridiculous.

However, even when public high school students can independently use the tool at home to learn more about it, they prioritize it for completing their homework. As AI abilities improve exponentially, all students need some basic knowledge to grasp this transformative tech; otherwise, this divergence will continue to widen.

I was able to help survey 600+ students from seven different schools within three other districts to get a sense of their perspectives on AI because they are the generation inheriting this technology. Additionally, gathering data across different schools and valleys allows responses based on diverse backgrounds and environments.

Here are their responses:

Lack of AI Familiarity

An overwhelming 45% of students were only familiar with ChatGPT while 33% of students don't know of any AI tools.

Lack of AI Usage

An overwhelming 66% of students rarely use AI in daily life and 51% of students don't use AI in schoolwork at all.

English the Primary Use of AI

For students that even use AI, 33% chose that English is the subject that they utilize AI tools to assist in their learning.

I wanted to learn more about students' insights regarding their pathways and the educational opportunities and policies present at our public high schools.

One additional question I asked was:

"How has your school experience helped you discover or pursue your passions and interests?"

Responses #1

"I enjoyed the freedom in my art class where I would create pieces through creative processes, but other than that school has not really helped me discover a passion. I usually found passion outside of school"

"School in my eyes has a set-curriculum that is locked into place and cannot be moved. Whether your major is similar or different to one another, the curriculum, is the exact same."

I also asked some other students to elaborate further on one question that was mentioned in the data set:

"What are your thoughts on teachers prohibiting the use of AI tools?"

Responses #2

It’s stupid, this is the future why drag us into the past

"Bullsh*t. I think AI allows students to learn and analyze the topic even deeper."

This is where this project all began; I learned various new things over time. I understood the different types of math utilized in Artificial Intelligence and how it works to answer a question. I also learned about the differences between deep learning and machine learning.

•  https://www.ibm.com/topics/artificial-intelligence 
•  https://www.forbes.com/advisor/in/business/software/what-is-ai/ 
•  https://www.youtube.com/watch?v=RzkD_rTEBYs 
•  https://www.youtube.com/watch?v=ad79nYk2keg 
•  https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence 
•  https://www.techtarget.com/searchenterpriseai/tip/The-history-of-artificial-intelligence-Complete-AI-timeline 
•  https://youtu.be/056v4OxKwlI?si=svbNdieYy6lGiuI7 
•  https://www.forbes.com/sites/forbesagencycouncil/2023/09/11/the-future-of-artificial-intelligence-predictions-and-trends/?sh=415019b23938 
•  https://www.bloomberg.com/company/press/generative-ai-to-become-a-1-3-trillion-market-by-2032-research-finds/ 
•  https://www.forbes.com/sites/jackkelly/2023/05/01/new-era-of-turbulence-the-world-economic-forum-predicts-25-of-jobs-will-change-over-the-next-five-years/?sh=18ac757b1775 
•  https://www.weforum.org/press/2020/10/recession-and-automation-changes-our-future-of-work-but-there-are-jobs-coming-report-says-52c5162fce/ 
•  https://youtu.be/UIZAiXYceBI?si=wSN5QbDcFpAMP_C6 
•  https://www.techtarget.com/searchdatamanagement/definition/data-preprocessing 
•  https://www.analyticsvidhya.com/blog/2019/10/mathematics-behind-machine-learning/ 
•  https://www.khanacademy.org/math/linear-algebra/vectors-and-spaces/vectors/v/vector-introduction-linear-algebra#:~:text=A%20vector%20is%20a%20quantity%20or%20phenomenon%20that%20has%20two,%2C%20electromagnetic%20fields%2C%20and%20weight. 
•  https://www.khanacademy.org/math/precalculus/x9e81a4f98389efdf:matrices/x9e81a4f98389efdf:mat-intro/a/intro-to-matrices 
•  https://www.quora.com/What-math-is-used-in-machine-learning-artificial-intelligence-research 
•  https://www.linkedin.com/pulse/crucial-role-mathematics-artificial-intelligence-shila-kishore#:~:text=Mathematics%20serves%20as%20the%20backbone,for%20developing%20machine%20learning%20algorithms. 
•  https://machinelearningmastery.com/a-gentle-introduction-to-multivariate-calculus/ 
•  https://www.trendhunter.com/trends/dora-ai 
• Claude AI
• Chat GPT-4
• beautiful.ai
• Youtube