A Human's Guide To Detecting AI Generated Content

the alignment problem: creatives and content

Another body formed in this cause is the United Nations' Multistakeholder Advisory Body on Artificial Intelligence, conceived as a part of the Secretary-General's Roadmap for Digital Cooperation in 2020. Recognizing the duality of AI's potential for both good and bad, it underscores the necessity for heightened multi-stakeholder efforts in AI cooperation on a global scale. It is currently in formation and hopes to forward recommendations for the international governance of AI.

Together, these entities and their guidelines show a collective commitment to meld AI's progress with the tenets of transparency, accountability, fairness and the overarching umbrella of privacy.

Even major tech conglomerates like IBM are taking steps to acknowledge their responsibility in regulating AI's societal impact. They are actively displaying their ethical principles on their websites and in 2020, Forbes reported that IBM decided it would no longer sell general-purpose facial recognition technology.

("Why It Matters That IBM Abandoned Its Facial Recognition Technology" Forbes, June 18, 2020) 

This decision reflects their concerns about potential misuse and advocating for a broader dialogue on its appropriate use. Such initiatives address the ethical, legal, and societal implications of AI and promote best practices.

The role of lawmakers and policymakers in this context cannot be overstated. They are tasked with the duty of revisiting existing laws with an eye toward evolving them to accommodate the unique challenges presented by AI.❞

Quotations taken from and inspired by "The privacy paradox with AI by Gai Sher and Ariela Benchlouch"

TLDR? There are many, many, many steps taken around trying to ensure that AI is used for the welfare of humanity rather than being abused. From laws to communities and several guidelines — they're doing their jobs but, are you?

To add, you might be curious about—

What data does AI even use?

- Publicly available data: Websites, books, academic papers, etc.

- Licensed datasets: Purchased or obtained through agreements.

- User-generated content: Social media posts, forums, etc.

- Specialized datasets: Created for specific training purposes.

How is it handled?

- Data cleaning: Removing personal identifiers and inappropriate content.

- Aggregation: Combining data from multiple sources.

- Anonymization: Stripping personally identifiable information.

- Encryption: Protecting data during storage and transfer.

What about privacy?

- Consent: Ensuring data is used with proper permissions.

- Compliance: Adhering to regulations like GDPR, CCPA.

- Transparency: Disclosing data usage practices.

- Data minimization: Using only necessary data.

Progress vs. Privacy - Where's the balance?

- Implement Federated learning: Train models without centralizing data.

- Differential privacy: Add noise to data to protect individual privacy.

- Synthetic data: Use artificially generated data for training.

Should we panic?

- There are valid reasons to be cautious about data privacy

- Many organizations are working to address these issues

- Users can take steps to protect their own data

Here's some things you can do:

- Be aware of what data you share online.

- Read privacy policies of services you use.

- Use privacy settings and tools available to you.

- Support initiatives and regulations that protect data privacy.

By approaching AI as a tool to enhance rather than replace human creativity, we can harness its power while mitigating potential risks and maintaining the unique value of human insight and expression in content creation.

And that's for now. We'll finally getting into how to detect AI writing from human writing, some tell-tale signs and all, I'd be mostly referencing my work, but if I reference the work of someone else, I'd make sure to credit them 🙌

Toodles!

Sara.