On the flipside of the many exciting AI innovations of the past few years we find a wide range of known and emerging risks: algorithmic bias, privacy concerns, and copyright infringement come to mind. That’s before we even begin to approach macro-level social problems, like the chance that millions of jobs might become obsolete in the not-too-distant future.
Data and ML professionals have been working hard to raise awareness of these concerns, and to come up with workable solutions that aim to balance technical progress with fair and responsible practices. It’s likely too early to tell just how successful they—and all of us—will be in threading that particularly fine needle. Still, it’s crucial to stay informed about the contours of these conversations if we ever hope to effect positive change in our professional communities (and beyond).
Our highlights this week tackle thorny questions around AI—from regulation to technical guardrails—with clarity and pragmatism. Whether you’re new to this topic or have been engaged with it for a while, we think these articles are worth your time.
Photo by mostafa meraji on Unsplash
Looking for excellent reads on other topics? You can’t go wrong with any of these top-notch articles:
Thank you for supporting the work of our authors! If you enjoy the articles you read on TDS, consider becoming a Medium member — it unlocks our entire archive (and every other post on Medium, too).
The Practical Side of Navigating AI Risks was originally published in Towards Data Science on Medium, where people are continuing the conversation by highlighting and responding to this story.
Data and ML professionals have been working hard to raise awareness of these concerns, and to come up with workable solutions that aim to balance technical progress with fair and responsible practices. It’s likely too early to tell just how successful they—and all of us—will be in threading that particularly fine needle. Still, it’s crucial to stay informed about the contours of these conversations if we ever hope to effect positive change in our professional communities (and beyond).
Our highlights this week tackle thorny questions around AI—from regulation to technical guardrails—with clarity and pragmatism. Whether you’re new to this topic or have been engaged with it for a while, we think these articles are worth your time.
- Legal and Ethical Perspectives on Generative AI
For an accessible primer on the interconnected issues gen-AI tools bring in their wake, Olivia Tanuwidjaja’s recent overview is a great choice: it offers just enough detail to orient you around this complex topic, and provides helpful resources for you to expand your knowledge of the areas you care about the most. - The Case Against AI Regulation Makes No Sense
The European Union’s AI Act is often touted as the most serious attempt (so far) to regulate the development and implementation of AI products; Adrien Book unpacks its most salient features, reflects on what it might still be missing, and advocates for more jurisdictions to think seriously—and proactively—about similar legislative initiatives.
Photo by mostafa meraji on Unsplash
- The Next Step is Responsible AI. How Do We Get There?
For a practical approach to responsible and ethical AI, Erdogan Taskesen proposes a 6-step roadmap that teams and organizations can adapt to their own needs. It serves as an important reminder that individual practitioners have agency, and they can leverage it to shape practices and choices in the process of building ML-based products. - OpenAI’s Web Crawler and FTC Missteps
The debate around copyright, artists’ work, and the way LLMs and image-generation models are trained has never been more contentious. Viggy Balagopalakrishnan provides a useful snapshot of the current stalemate by focusing on recent news from OpenAI and the challenges the FTC (Federal Trade Commission) faces in its attempts to regulate well-funded tech companies. - Safeguarding LLMs with Guardrails
Controlling the reach, scope, and effects of AI tools is important on a micro-local level, too: if you’re working on a large language model integration, for example, you definitely don’t want it to spew offensive language or to insist that an hallucination is factual. Aparna Dhinakaran and Hakan Tekgul share a hands-on guide to open-source tools that allow developers to enforce strict parameters on model outputs.
Looking for excellent reads on other topics? You can’t go wrong with any of these top-notch articles:
- If you’re taking your first steps in environmental data science, Caroline Arnold’s accessible introduction is a great place to start.
- How did machine learning evolve from encodings to embeddings? Mina Ghashami’s explainer is both comprehensive and beginner-friendly.
- For a fun—and enlightening—tutorial on transformers’ strategizing abilities, don’t miss Charlie O'Neill’s walkthrough of building a tic-tac-toe simulator.
- To help you decode job descriptions and what they actually entail in real life, Stephanie Kirmer outlines a few key archetypes of data-science roles.
- In the market for a math deep dive? You’ll absolutely want to bookmark Gabriel de Longeaux’s detailed exploration of the two-envelope paradox.
- Ending on a practical note, we highly recommend Mariya Mansurova’s recent step-by-step guide to using BERTopic for advanced topic modeling.
Thank you for supporting the work of our authors! If you enjoy the articles you read on TDS, consider becoming a Medium member — it unlocks our entire archive (and every other post on Medium, too).
The Practical Side of Navigating AI Risks was originally published in Towards Data Science on Medium, where people are continuing the conversation by highlighting and responding to this story.
