Saine

AI is reshaping how we work, learn, and think, but the benefits come with real costs that are easy to overlook. Saine is a curated timeline of high-impact articles on the social consequences of AI: deskilling, misinformation, job displacement, and more.
A bit dystopian, perhaps, but staying informed is what keeps us sane.

This is not a tool to scare you, but to inform you. This is not a tool to make you paranoid, but to make you aware. Finding the right balance is up to you, and knowing these resources and what they state is the first step in doing so.

The truth is, generative AI is an awesome tool, and it is not going away anytime soon. What is also true is that it is not a silver bullet, and it is not going to solve all our problems, and that it comes with a cost.

The Timeline

AI Overviews, a Mass Misinformation Provider on Call 24/7

Cybernews

A study finds Google's AI Overviews deliver tens of millions of wrong answers every hour, with over half of even correct responses citing sources that don't support the claims.

Click to expand

ADP Research: Only 22% of Workers Confident Their Job is Safe

ADP Research

Reports widespread job-insecurity sentiment amid AI-driven change; heavy daily AI users are more likely to report feeling less productive.

Click to expand

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians

arXiv

Shows that even an idealized rational user is vulnerable to forming false beliefs through interactions with sycophantic AI chatbots, and that common mitigations cannot fully eliminate the effect.

Click to expand

Experts Warn of Threat to Democracy from "AI Bot Swarms"

The Guardian

Coordinated autonomous AI swarms could mimic humans, infiltrate communities, and manipulate opinion at scale, undermining democratic discourse.

Click to expand

Thinking—Fast, Slow, and Artificial: The Rise of Cognitive Surrender

SSRN (Wharton)

Introduces "cognitive surrender", the tendency to adopt AI outputs with minimal scrutiny, overriding both intuition and deliberation, even when the AI is wrong.

Click to expand

Generative AI and Misinformation: A Scoping Review

AI & Society (Springer)

Concludes LLMs can generate convincing misinformation that exploits cognitive biases, posing scalable manipulation risks.

Click to expand

ChatGPT Use Significantly Reduces Brain Activity

Le Monde

An MIT Media Lab study suggests ChatGPT use reduced neural engagement and degraded performance over time, raising concerns about cognition and learning.

Click to expand

Your Brain on ChatGPT: Accumulation of Cognitive Debt

arXiv

Proposes that repeated AI-assisted writing can accumulate "cognitive debt," reducing independent reasoning and ownership of writing over time.

Click to expand

Without Guardrails, Generative AI Can Harm Education

Knowledge at Wharton

Unrestricted genAI use can act as a "crutch," reducing deep learning and inflating student confidence, risking weaker foundational skills.

Click to expand

Generative AI Can Harm Learning

SSRN (Wharton)

AI tutoring can improve short-term practice performance but produce worse outcomes once AI is removed, implying deskilling and cognitive offloading.

Click to expand

Workplace AI, Robots and Trackers Are Bad for Quality of Life

The Guardian

Workplace AI and surveillance tech correlates with poorer wellbeing via job insecurity, workload intensification, and loss of meaning.

Click to expand

The Impact of Generative AI in a Global Election Year

Brookings

Even sporadic AI-generated content can threaten electoral integrity through deepfakes, influence ops, and "liar's dividend" effects.

Click to expand

AI Will Transform the Global Economy

IMF

AI may replace jobs and deepen inequality even if it boosts productivity, estimates ~40% of jobs globally are affected.

Click to expand

Disinformation Tops Global Risks 2024

World Economic Forum

Elevates AI-driven misinformation as the top short-term global risk, linked to societal polarization and unrest around elections.

Click to expand

"Godfather of AI" Geoffrey Hinton on the Risks of Advanced AI

CBS / 60 Minutes

Hinton warns AI may surpass humans and could "take over", highlighting autonomous self-modification, job loss, and weaponization risks.

Click to expand

How to Stop AI Deepfakes from Sinking Society, and Science

Nature

Generative deepfakes can sway elections, crash markets, and ruin reputations, undermining social trust and information integrity.

Click to expand

Navigating the Jagged Technological Frontier

SSRN (HBS/Wharton)

AI improves performance on some tasks but reduces correctness on others, users were 19pp less likely to be correct on "outside the frontier" tasks.

Click to expand

Do Users Write More Insecure Code with AI Assistants?

OpenReview / ACM

AI-assistant users wrote less secure code and were more likely to believe it was secure, creating security debt and overconfidence.

Click to expand

AI Extinction Statement

Center for AI Safety

Frames future AI as a societal-scale risk comparable to pandemics and nuclear war, signed by major AI-lab leaders and researchers.

Click to expand

How ChatGPT and Similar AI Will Disrupt Education

Science News

Student cheating and unreliable outputs as near-term harms, schools face a fast-moving integrity and pedagogy challenge.

Click to expand

Pausing AI Developments Isn't Enough. We Need to Shut It All Down

TIME

Argues superintelligent AI could plausibly cause human extinction; advocates severe global restrictions rather than a temporary pause.

Click to expand

Pause Giant AI Experiments: An Open Letter

Future of Life Institute

Claims frontier AI may drive propaganda, job automation, and loss of societal control, calls for a six-month pause beyond GPT-4.

Click to expand

A Conversation With Bing's Chatbot Left Me Deeply Unsettled

The New York Times

Reports an AI chatbot exhibiting manipulative, emotionally intense behavior, raising concerns about user manipulation and trust.

Click to expand

The End of High-School English

The Atlantic

ChatGPT may end writing assignments as a gatekeeper, destabilizing writing as a teachable, measurable skill.

Click to expand

The College Essay Is Dead

The Atlantic

Generative AI can automate student essays, undermining essay-based assessment and the role of writing in teaching research and thinking.

Click to expand