Local Resource: Together We AI

This past summer (August 2025), we had the opportunity to collaborate with Together We AI, a non-profit started by a high schooler in the South Bay. They graciously agreed to run three virtual sessions for us. Originally the sessions were designed for middle/high schoolers but so many parents wanted to learn as well that we ended up opening the session to all ages.  

Session 1: Overview of AI

We learned that AI is defined as a software system that makes the best/most logical decision to reach a goal. The implementations we see nowadays use training data to select what is "most logical", but AI itself has a history spanning back to the 1950s, using algorithms to determine what is best.

Image source: Waymo

AI is now embedded in many systems, e.g. self driving cars. As such, parents are concerned about careers in the age of AI. Indeed, routine jobs will be at risk, but there will also be new jobs, e.g. AI oversight, ethics. Existing workers will often need to understand how to use AI in your work, e.g. Canva AI tool for graphic designers

Further reading/AI sites to try: 

• https://medium.com/@contact.togetherweai/an-overview-of-ai-for-beginners-0c7ddec43177
• https://quickdraw.withgoogle.com/#
• https://experiments.withgoogle.com/ai/ai-duet/view/
• https://www.semantle.com/ 

Session 2: Generative AI and AI in Healthcare

In this session, we learned that generative AI is a subset of AI which creates content based on a user's prompt. It requires massive amounts of data to train so that it recognizes patterns, and is "tuned" to evaluate if its generated output matches what users want. An example is Canva's AI Image Generator (link does not require a Canva account).  Generative AI is popular, but also comes with problems. such as hallucinations and plagiarism.

AI has also been deployed in healthcare, e.g. as a virtual health care assistant to schedule appointments or assess symptoms, or in disease detection (E.g. detecting Alzheimer's in MRI scans). Its issues include serious consequences for mistakes, ethical dilemmas and widening the health gap globally.  

Session 3: Ethics in AI

Ethical questions arise when people lack agreement on moral beliefs and priorities. AI has added fuel to the fire because of its transformative power. Examples include:

• Discrimination: Biased/unrepresentative training data = biased model, perpetuate stereotypes
• Privacy: Enormous amounts of sensitive data used for training, security and misuse
• Blurring of reality: Deep fakes are increasingly difficult to detect and used in misinformation.
• Environment damage: High water and power use, e.g. a ChatGPT conversation uses 1 plastic bottle's worth of water

To address these issues, developers should test AI models vigorously with a wide data set and diverse ideas, e.g. Generative Adversarial Networks/GANs. It's harder for consumers to detect biased AI models, but being aware they exist and advocating for better models is a start. The European Union AI Act gives some guidelines for using AI ethically, and companies are also creating jobs, e.g. AI Ethicist, to focus on these areas. 

We were also asked to determine if some photos were deep fakes, and learned some tips for recognizing them:

• Shadows: AI shadows often are unrealistic/defy physics
• Focus: AI makes highly focused/good quality pictures but real life photos often aren't that way, e.g. clear foreground and blurry background, super smooth skin
• Details: AI models overcompensate by adding details not in real life, e.g. multiple fold lines on a coat, or getting details wrong, e.g. floating pen in coat pocket

Conclusion/Further Reading

We are grateful for the time Together We AI took to present to us, and especially appreciated the opportunity to ask questions. They will continue to present webinars and their website is the best place to stay updated.  Homeschoolers may also want to follow up with:

• Stanford Digital Group: Not directly related to AI, but this series of lessons teaches students to evaluate online sources through cross checking (reading laterally). The Senior Prank lesson is a fun, standalone exercise to warm up.   
• Elon University has published a Student Guide to AI , exploring ethical guidelines for academic use of AI