The use of punch cards to write code came with various disadvantages including but not limited to:

• Each card had very limited storage which required the use of multiple cards for a single program.
• It was incredibly difficult to reconstruct a program if any accidents occurred.
• In order to fix a simple mistake you would need to rewrite the whole card.
• They can’t be reused and are bulky to handle; therefore, they were expensive to produce and store.
• Easily damaged, lost, or stolen which results in a loss of data.

The rise of programming languages provided a solution to these problems. Ultimately, the purpose of programming languages was making the programming process more efficient by making it easier for humans to write and store their code.

There are hundreds of problems that we need to code solutions for; therefore, we need various programming languages, each designed to specifically address each of these problems. For example, when you need to manipulate and analyze data you would use SQL and R, languages designed specifically to handle analyzing large amounts of data. On the other hand, in web development you would choose a versatile language that can be used for interactivity and animations such as Javascript.

Python’s main drawback is its execution speed. Since python isn’t pre-compiled, its execution time is higher relative to other languages. This drawback is emphasized in tasks where you need low-latency and quick execution (such as in animations and video games). It would be better if python’s interpreter can be optimized for faster code execution.

• The first thing I would need to define is the purpose of the programming language. i.e. What problems do I want to solve with this new language? and What features do other languages lack that I want to include in my own language?
• The second thing I would need to design is the language’s structure and syntax
• Finally, I would program an interpreter that would allow the computer to understand my language and execute my code.

A decision problem is a task where a given input is evaluated based on whether or not a property is present. The solution is always one of two outputs, either true or false.

A decision problem is decidable when it has a solution that can be calculated by an algorithm in a finite amount of time

Class P and NP are used to classify a problem's complexity. The P class is the class of problems that can be solved using deterministiic machines(a machine that returns the same output for the same input such as our computers) in polynomial time. P problems are usually easy to compute. NP problems ,on the other hand, are ones that can only be solved by non-deterministic machines in polynomial time. This makes them hard to compute but easy to verify.

P is the set of all problems whose solution can be computed easily. NP is the set of all problems whose solutions can be easily verified. NP vs P asks if NP problem's solutions can be easily verfied then can they also be calculated in polynomial time by deterministic machines.

One Million dollars richer

HCI is a field that aims to improve the interactions between humans and computers/machines. It combines various disciplines including computer science, psychology, design, and social science

A usable system is one that is easy to use and intuitive to navigate. On the other hand, a useful system is one that actually serves a meaningful function/purpose. For example, let's say you want to build a volume controller. You can decide to make a single button which you click and it gives you a random volume. This system is usable. All you need to do is click a button in order to change the volume. However, this system is utterly useless as it just gives random volumes and gives you no control over the actual value, hence rendering it useless.

One everyday interface that is confusing to use is the dorm's washing machine's keypad. A person using the washing machine would want to pick an appropriate cycle for the type of clothes they are washing; however, a new resident is hindered by the buttons that are only labeled with random symbols. An improvement would be to properly label the different buttons to indicate the appropriate cycles.

A prototype is an initial model that is made to test the product's functionality and design before its released to the masses. Testing a prototype with users provides the designer with feedback that would allow them to fix any problems that they might have not noticed and ensure that the launched product is actually functional and useful for potential buyers. Asking "Do you like it?" often prompts a yes or no answer which doesn't pinpoint areas of strengths nor of weaknesses.

A voice control feature on mobile devices. This feature would benefit the visually-impaired by allowing them to utilize their phones without needing to see the interface. In order to test whether this is a beneficial feature or not, I would utilize the accessibility evaluation tools + user feedback through surveys.

Once a problem becomes too big for one computer, we can divide the tasks on multiple computers connected through a network.

No. Yes the total processing power increases by 1000 times but other elements such as communication & coordination between computers, and the sequential nature of some tasks are introduced and managing these factors reduces the overall performance increase.

Yes, today there is a system known as Raft. Raft is designed to deal with situations where there is failure by reaching consensus based on the majority not on every individual computer. A malicious node is combated by Byzantine Fault Tolerance Protocols which also select the decision based on the majority but they recognize that some computers might be malicious and account for that.

In the cloud. Most of the computation occurs in a remote data centers. This ensures that you can access your information from any location, device, and time.

If we can have one million computers behaving as one machine we would be able to develop perfect modeling systems that would allow us to improve various fields such as weather forecasting and prediction.

A specific form of data that an AI model can understand and learn from. This includes text, audio, and images.

In the context of AI, multimodal learning refers to models capable of processing information from various modalities or types of data

Multimodal learning is used whenever you need to manipulate different forms of data for tasks such as image recognition and diagnosis. One real world application is Google Gemini that can take audio, image, text,and vidoe inputs and produce any of those forms as output.

Embodied AI is when physical systems are integrated with AI models which enables them to perceive and interact with the real world.

Embodied AI depends on multimodal learning. In order for an embodied AI to interact with the world it must be able to process different types of inputs such as audio, visual, and even sensor input. Thus, multimodal learning is required to enable Embodied AI.

A type of technology that aims to mimic human learning, problem-solving, and ablity to think.

Computer Vision, Natural Language Processing, and Generative AI

The concepts behind AI have been around for decades; however, we didn't possess the computing power and massive data sets required to bring these concepts into reality. The development of faster processors, the increase in available data sets, and the rise of cloud computing solved these problems and allowed researchers to develop models that can recognize images, understand speech, and perform complicated tasks. With that came the boom of AI.

Robots are used within the surgical field, the automotive manufacturing sector, and the food production sector. Each of these applications require precision and efficiency which these robots excel at.

1. Navigating moving obstacles and crowds 2. Managing its power and battery and any potential technical malfunctions 3. Tracking its position within the facility and navigating its different levels