Charlie & Paul MIT Museum Project

MIT Museum

Exploring Synthetic Biology with Paul

Exploring the Capabilities and Pitfalls of AI with Charlie

View Paul's Reflection!

View Charlie's Reflection!

Commuted from Bentley University to the MIT Museum for a tour on Artificial Intelligence and individual exploration

Research On Masked Bias

Gaps in AI

Recognizing Disparities

How Did Scientists Do This?

Scent Diffuser

<--- Exhibit Description

Paul's Reflection

Return to Home Page

• Deepfake technology requires us to filter what we see with healthy skepticism
• Rapidly growing tech needs to be regulated with ethical rules simultaneously
• Non-invasive wearable tech can help people with speech disorders

Alter Ego MIT Webpage

Learn more about Alter Ego

Charlie's Reflection

Return to Home Page

• AI can be as destructive as it is helpful
• If AI isnt carefully monitoured and checked over it can develope into a potentially harmful resource.
• The MIT also contained many inventions and prototypes that included but were not limited to things that decrease fossil fuel emissions

Learn more about Lithium Air Battery's

Does Training Data Matter

AI Disparities

https://arxiv.org/html/2312.10833v2

In this study we can learn that diversifying data sets and training AI systems on more data can help eliminate bias and lower equity concerns regarding AI.

In the study an important example included the research on discrimination AI can have against women. "AI takes the form of influential automatic scoring systems that reproduce societal stereotypes and gaps". Although sometimes its unintetional mechanical bias can lead to discriminatory outcomes.

In conclusion its necassary to include diversity in machine learning and training data so that things like gender bias are not prevelant and so AI doesnt learn and create disparities.

What's the Purpose?

Source

The purpose of this scientific project was not de-extinction. In fact, scientists are not able to pinpoint th exact strength or amount of scent the flowers produced. Instead, by providing a visual video of what the flower might've looked like in its natural habitat before it was destroyed by humans and re-creating its smell, it helps the audience experience what has been lost.

Masked Bias

https://news.mit.edu/2018/study-finds-gender-skin-type-bias-artificial-intelligence-systems-0212

In the photo you can see an example of what the AI moduel recognized as a human face. This is critical to the study becasue it was found that commercially released facial-analysis programs had skin-type and gender bias.

Study finds gender and skin-type bias in commercial artificial-intelligence systems

In a chance discovery Dr. Buolamwini recognized that the AI was unable to work with darker skinned individuals. After further research she found that the AI had an error rate of over 34 percent for darker-skinned women while it was never worse than 0.8 for light-skinned men.

This shows the implications of AI and how they can have signifigant biases. If left unchecked certian AI similar to this could end up dangerous and cause more harm than good. If a facial recognition program like this was released for crime matching it could lead to dangerous misclassificationsIn general this study shows that AI can be both beneficial in the sense that it can help with facial recognition but dangerous in the sense that it has both gender and race bias.

• Ginkgo scientists create yeast and bacteria that secrete chemicals
• These chemicals can also produce smell molecules
• Extract DNA samples from historic plant samples
• Predict gene sequences which contain fragrance producing enzymes

How Did Scientists Do This?

Source