Fundamentals of Ai

What is Miro?

Miro is a website that's made especially for managing and assisting with team communications. Miro really impressed me and happened to be quite useful. Working in a team is made a lot easier by its easy-to-use interface, which also automatically saves all of our lecture-related work so we can look over our notes anytime needed.

What is AI and how can it generate media with prompts

The term artificial intelligence (AI) describes technology that makes it possible for computers to replicate human intelligence and the ability to solve problems. Artificial Intelligence essentially allows computers to learn and solve problems like humans. Using its learning algorithms, AI uses prompts as cues to produce art that combines creativity, precision, and randomness. It can create entirely original aesthetics or replicate art forms that already exist whilst following the prompt given.

Why is AI important in todays day and age?

AI is becoming increasingly important in today's world. It functions as a kind of intelligent sidekick for humans. Artificial Intelligence can perform all the labor-intensive tasks. It's comparable to having a personal assistant who never grumbles about working extra hours.
It can automate various tasks, ranging from categorising cat photos to forecasting stock market strategies. It's all part of AI's job. It also helps make everything more personalised to you based on your exact preferences and needs for example Netflix. It gives you "recommended" shows all based on your binge history. AI also helps with things such as wether forecasts and epidemic outbreak predictions, making the job much easier for not only regular people but also scientists.

Is AI programmed with ethics and morals?

To ensure responsible AI development and use, a set of fundamental principles are involved in AI ethics. Part of these fundamental principles are:

Safety: Making certain AI systems are humane, safe, and secure.

Fairness: Resolving prejudices and guaranteeing just results.

Transparency: Providing an explanation for AI decisions and processes.

Although there are many safety and precaution measures put in place, AI systems have the potential to violate human rights, cause prejudice, and increase climate change. As AI technologies advance, existing gaps in those safety measures can even increase bias and may even hurt marginalised communities.

Group discussion: Bias and Fairness using AI in the real world.

Group 1 - "Bias and Fairness: Facial Recognition Misidentification"

I was paired up with Cassie to form a group and to discuss our opinions on a case study given to us. We were given a study about Robert Williams, a Black man from Detroit who was wrongfully arrested because of AI facial recognition.

We came to a few important conclusions after reading and evaluating the article. First of all, the responsibility for the AI’s behavior lies more with the people who trained it and provided an input into it. The AI will unavoidably generate biassed results if it is fed racist or biassed data. Thus, blame should be directed at the humans, not the mindless machine. Secondly, AI should not be tasked with identifying suspects. While it can assist in evidence gathering and analysis, it should stay within the capabilities of human judgement to decide wether or not someone has done a certain crime and if they are the real suspect or not.

History of Chatbots:

Large Language Models:

Large Language Models:

- A large language model (LLM) is an advanced AI system trained on massive amounts of text data to understand, interpret, and generate human-like language.

- Composed of a parameters file and code file to run the parameters.

Neural Networks:

Neural networks, especially in large language models like ChatGPT, are like digital brain networks. Imagine they’re inspired by how our human brains work—lots of interconnected neurons firing away.

The thing is that AI:

Spots Patterns: These networks are super good at spotting patterns in massive data piles.
Learning: They learn by example, just like we do. Show 'em some text, and they get smarter.
Layers of Nodes: Picture layers of tiny interconnected nodes (kinda like digital neurons). These nodes process info and make predictions based on what they see.

Best model of AI (accessible to public):

- GPT-4: Expert in artistic pursuits, including poetry. It a much more human-like voice than other LLMs.
But because of its huge set of training data, it often gives false information.

How to put together a prompt AI that is effective:
Provide a clear instruction: To ensure that the AI understands your request, please be as specific and clear as possible.
Keep it brief: Adopt succinct, straightforward language.
Give the background: To help the AI understand context, include pertinent details.
Verify the grammar: Check that the grammar in your prompt is correct.
Try out some outputs: Try a variety of prompts to find the best response.
Discover the model's details: To varying prompts, LLMs will react in unique ways.
Adapt the AI: For more precise and customised results, use your data to fine-tune the model when it is available.

Image Generation:

What are the current best image-gen tools?

Midjourney:

Excellent Image Quality: Midjourney often produces images with a great deal of to detail, style, and even an artistic quality.
Creative Control: It provides a number of settings and customisation choices so you may fine-tune and direct the process of creating images.
Rich Environment for Sharing Creations: The vibrant and supportive Midjourney community offers encouragement, inspiration, and support.