GEMINI @ PRATT

Share this website from your phone

Understanding Gen AI

Gen AI Progress: Generative AI's journey to its current capabilities, exemplified by models like Gemini, spans decades. Its origins trace back to the foundational concepts of artificial intelligence and machine learning in the mid-20th century, with early explorations into neural networks and simple chatbots like ELIZA in the 1960s. Progress accelerated significantly with increases in computing power, the availability of massive datasets, and breakthroughs in deep learning algorithms. Key milestones include the development of Generative Adversarial Networks (GANs) in 2014, which greatly improved the realism of generated images and media, and the introduction of the Transformer architecture in 2017. Transformers revolutionized how AI processes language, leading directly to the powerful Large Language Models (LLMs) we see today, enabling AI to not just understand information but to generate novel, coherent, and contextually relevant content like text, images, code, and more.

Types of AI with sample use cases (Expand)

Machine Learning (ML)

What it is: A technique used to train computers to learn from data and make decisions or predictions.

Used for: Many tasks like fraud detection, recommendations, and more.

It's the engine behind most modern AI systems, including Narrow and Generative AI.

Email Spam Filtering
How it works: Machine learning models analyze the content, sender, and structure of emails to classify them as "spam" or "not spam."
Impact: Helps keep your inbox clean by learning from past flagged messages and adapting to new spam techniques.
Recommendation Systems
Example: Netflix, YouTube, and Spotify.
How it works: Algorithms learn your preferences based on past behavior (watching, clicking, liking) and suggest personalized content.
Impact: Increases user engagement and satisfaction.
Fraud Detection
Example: Credit card companies and banks.
How it works: Machine learning models learn patterns of legitimate transactions and flag unusual or suspicious activity.
Impact: Reduces financial losses and protects users from fraud.
Voice Assistants
Example: Siri, Alexa, Google Assistant
How it works: Natural Language Processing (NLP) and machine learning are used to understand spoken language, interpret intent, and respond appropriately.
Impact: Enables hands-free control of devices and instant access to information.
Medical Diagnosis
Example: Detecting diseases from X-rays or MRIs.
How it works: Models are trained on large datasets of medical images to identify abnormalities like tumors or fractures.
Impact: Supports doctors with faster and more accurate diagnoses.

Narrow AI (Weak AI)

What it is: A type of artificial intelligence designed to perform one specific task with a high level of accuracy.

Used for: Tasks like facial recognition, language translation, spam filtering, or navigation.

It’s the most common form of AI in use today—focused, efficient, but limited to its programmed domain.

Google Maps Navigation
What it does: Calculates the best route based on traffic, road conditions, and past travel data.
Why it’s Narrow AI: It’s smart at giving directions but can’t do anything else.
Face Recognition in Smartphones
What it does: Unlocks your phone using facial features.
Why it’s Narrow AI: It recognizes faces only—it can't recognize emotions, actions, or anything outside its training.
Chatbots for Customer Support
Example: Bank or airline chat assistants.
What it does: Answers common questions and helps with simple tasks like checking a balance or booking a ticket.
Why it’s Narrow AI: It only knows how to handle predefined support topics, not general conversation.
Autonomous Roomba Vacuum
What it does: Navigates your home to clean floors.
Why it’s Narrow AI: It can’t cook, talk, or understand broader commands—just clean.
License Plate Recognition (LPR) Systems
What it does: Detects and reads vehicle license plates for tolls, parking access, or law enforcement.
Why it's Narrow AI: Focused only on identifying characters in a specific format—nothing else

Generative AI

What it is: A type of AI that can create new content—such as text, images, code, audio, or video—by learning patterns from existing data.

Used for: Writing responses, generating artwork, composing music, coding, designing presentations, and more.

Tools like Google Gemini use Generative AI to enhance productivity, creativity, and automation across Gmail, Docs, Sheets, Slides, and other Workspace apps.

Generating Marketing & Sales Content.
How: Gemini could automatically draft emails for sales outreach, create initial versions of marketing brochures highlighting Pratt's 100% recycled packaging solutions, generate social media posts about sustainability achievements, or write product descriptions for new packaging designs. Why it's Generative AI: It creates new, original text content (emails, posts, descriptions) based on prompts and instructions provided by the Pratt team, rather than just retrieving existing information.
Summarizing Operational & Sustainability Data.
How: Instead of manually sifting through spreadsheets, Gemini could analyze large datasets (e.g., plant production efficiency, logistics performance, energy usage across facilities) and automatically generate easy-to-understand natural language summaries, identify key trends, and create draft reports for management or sustainability disclosures.
Why it's Gen.AI? It synthesizes complex data and generates concise, human-readable summaries and insights in text format, effectively creating a new representation of the information.
Powering Customer Service & Internal FAQs.
How: Gemini could power a chatbot on Pratt's website or internal employee portal. It could answer frequently asked questions about order status, product capabilities, recycling services, or internal HR policies by generating conversational, helpful responses based on Pratt's knowledge base.
Why it's Gen.AI? It generates human-like, contextual responses in a conversation, creating replies dynamically rather than pulling from a fixed list of pre-written answers.
Drafting Training Materials & Safety Guides.
How: Gemini could assist in creating initial drafts of training manuals for new machine operators, developing safety guides based on specific plant requirements or OSHA regulations, or outlining Standard Operating Procedures (SOPs) for manufacturing or recycling processes based on key input points.
Why it's Gen.AI? It generates structured instructional content, like manuals and guides, by organizing information and writing explanatory text based on the requirements provided.
Brainstorming Packaging Design Concepts.
How: Pratt's design teams could use Gemini to brainstorm ideas for new packaging structures or features. By providing criteria (e.g., "lightweight packaging for produce," "e-commerce box with easy opening," "sustainable alternative to plastic wrap"), Gemini could generate written descriptions of potential design concepts or innovative features to explore.
Why it's Gen.AI? It creates novel ideas and descriptive text for concepts that didn't previously exist, based on the creative prompts and constraints given.
Summarizing Meetings & Extracting Action Items.
How: After internal meetings (like production planning, sales reviews, or safety committee meetings), Gemini could process a transcript or recording and automatically generate a concise summary of the key discussion points, decisions made, and a list of specific action items assigned to different team members.
Why it's Gen.AI? It analyzes unstructured conversation and generates new, structured content in the form of summaries and action item lists, condensing and reformatting the information.