Artificial Intelligence and Future Markets

CMU 11-651 Course: Artificial Intelligence and Future Markets

Course Objective —

Familiarize students with AI techniques that are used in wide variety of application fields, learn to generate new ideas for AI products in a given field, review the business trajectories of 48 AI companies: who establishes them, how they are funded, and what their market impact has been, and develop presentation skills. This course is taught by Dr. Michael Shamos.

Task —

In a team of 4-5 students, I was involved in the research and presentation design for how AI is used in these randomly assigned fields: Identity Verification, Data Privacy, Firefighting, Personalized Medicine, Perfume, and Advertising.

Out of these, I gave a professional 30-minute presentation to teach the audience how AI is used in the Perfume industry. I explained how AI is currently used in the assigned field using three example applications, presented the specific AI techniques that were used and whether they were successful, and also proposed future AI applications in the given field. In addition to the AI techniques, I gave details about the formation and evolution of Firmenich, the world's largest privately-owned fragrance and taste company.

AI + Advertising —

AI is increasingly being used in the advertisement industry to improve targeting, personalization, and optimization of ad campaigns. AI can help advertisers identify and analyze consumer data to understand their interests, preferences, and behavior patterns. This data can then be used to create highly targeted and personalized ads, leading to better engagement and conversion rates. AI can also be used to optimize ad campaigns by analyzing real-time performance data and making adjustments to improve effectiveness.

In this presentation, these were the three applications my team focused on:

• Airbnb Real-Time Personalization Ads

• TikTok Ads Recommendation Ranking

• Personalized Generative Ads

AI + Data Privacy —

One application of AI in this field is the development of algorithms that can automatically detect and redact personally identifiable information (PII) from datasets. This can help ensure that sensitive data, such as names, addresses, and social security numbers, are not inadvertently shared or used inappropriately. Another application of AI in data privacy is in the development of privacy-preserving technologies, such as differential privacy. This involves adding noise to datasets to protect the privacy of individual users while still allowing meaningful analysis of the data.

In this presentation, these were the three applications my team focused on:

• Federated Learning with GBoard

• Voice De-identification

• Scalable Private Learning with PATE

AI + Firefighting —

One application of AI in firefighting is the use of drones equipped with AI-powered cameras and sensors to quickly detect and analyze the severity and spread of fires. AI-powered predictive models are also being developed to provide firefighters with real-time information about the likelihood and intensity of fires in different areas, allowing them to plan and allocate resources more effectively. In addition, AI-powered virtual reality simulations are being used to train firefighters in different firefighting scenarios and enhance their decision-making skills.

In this presentation, these were the three applications my team focused on:

• Training Data Synthesis from Simulation

• Real-Time Forecast of Tunnel Fires

• Simulation of Forest Fire Spread

AI + Identity Verification —

Various AI techniques such as facial recognition, biometric authentication, and document analysis are used for identity verification. Facial recognition technology can match an individual's face with their ID photo, while biometric authentication can use a person's unique physical characteristics like fingerprints or voiceprints to confirm their identity. Document analysis can be used to check the authenticity of identification documents like passports or driver's licenses. AI-powered identity verification systems are becoming increasingly popular in various industries, including finance, healthcare, and online platforms.

In this presentation, these were the three applications my team focused on:

• Keystroke Dynamics

• Facial Identification

• Voice Spectrography

AI + Perfume —

The use of AI in the perfume industry involves developing algorithms to analyze large datasets of scent molecules and consumer preferences, as well as creating computer simulations to predict how different scent combinations will be received. AI can also be used in the development of personalized fragrances, where individual customer preferences are analyzed and used to create unique scent formulations. Additionally, AI-powered scent sensors can be used in quality control and to detect counterfeit products.

In this presentation, these were the three applications my team focused on:

• Scent Generation

• Optimal Scent Generation: Concentration Variation

• Perfume Recommendation

AI + Personalized Medicine —

AI is being used to analyze large amounts of medical data, including patient medical records, genetic data, and clinical trial data, to identify patterns and correlations that can inform personalized treatment plans. AI is also being used to develop predictive models that can identify patients at high risk of developing certain conditions and to develop targeted therapies that are tailored to the specific genetic profile of individual patients.

In this presentation, these were the three applications my team focused on:

• Radiation Oncology: Multimodal Image Registration

• Deep Variant DNA Sequencing

• Precision Medicine for Chronic Diseases