What I read this week on AI - week 14
Articles on AI that caught my eye and can be linked back to internal audit
Last week, I read a range of insightful articles about AI—from demystifying how large language models “think” to innovative strategies in data security, critical reflections on AGI, transformative business cases, and even a fun quiz to test your smarts against AI. Each piece offers unique perspectives that can inform our approach to internal audit, risk management, and strategic governance.
Here’s what caught my attention this week:
Overview
How AI ‘thinks’: Researchers use neuroscience-inspired methods to peek inside LLMs’ decision-making.
Trapping AI in a Labyrinth: A new strategy to mislead unauthorized web crawlers with AI-generated decoys.
Rethinking Artificial General Intelligence: AI experts argue that current research paths may be misguided, driven by hype rather than robust progress.
AI-Driven Sales making the difference: A deep dive into how AI tools can transform the customer journey and drive new business.
Are You Smarter Than AI: A playful quiz to test your knowledge and see how you fare against the rapidly evolving capabilities of AI.
Anthropic Scientists Reveal How AI Actually ‘Thinks’
Anthropic researchers are pioneering methods inspired by neuroscience—using techniques like “circuit tracing” and “attribution graphs”—to map the inner workings of large language models. Their work reveals that these models may plan ahead or even work backward from a desired outcome, challenging the notion of AI as a complete black box.
For internal auditors, understanding how AI systems arrive at decisions is key. This research underscores the importance of model interpretability, which can lead to improved auditability and enhanced trust in AI-driven processes. It reminds us that even the most advanced systems benefit from transparency in decision-making.
Read more
(be sure to watch the video for quick understanding)
Cloudflare is Luring Web-Scraping Bots into an ‘AI Labyrinth’
Cloudflare’s new tool, AI Labyrinth, turns the tables on unauthorized web scrapers by directing them to AI-generated decoy pages. Instead of outright blocking these bots, the system confounds them, ultimately wasting their resources and safeguarding proprietary data.
This creative strategy highlights a crucial challenge: the insatiable appetite for data that AI systems possess. While Cloudflare’s method cleverly diverts bots, it also underscores concerns for websites that are increasingly at risk of having their data scraped. For risk management perspective, ensuring that the data ingested by AI systems is both accurate and secure is vital. This situation calls for robust data governance practices and proactive risk management strategies. At the same time, this approach highlights the possibilities of AI to be used to detect and counteract malicious behaviour.
Read more via The Verge
Announcement by Cloudflare
AI Experts Say We’re on the Wrong Path to Achieving Human-Like AI
A panel of AI researchers at the AAAI 2025 Presidential Panel warns that current approaches to AGI may be misdirected, driven more by hype than by sustainable progress. The report calls for a measured, safety-first approach, emphasizing collaboration, ethical governance, and a recalibration of expectations.
For internal audit professionals, this article highlights the need to critically assess AI investments and project outcomes. The gap between public hype and technical reality reinforces the importance of robust governance frameworks and realistic performance benchmarks when evaluating AI initiatives.
Read more via Gizmodo
Direct download the report ‘Future of AI Research’
How One Company Used AI to Broaden Its Customer Base
In this in-depth case study, SAP demonstrates how it transformed its sales strategy by integrating over 40 AI tools. By mapping the customer journey—from discovery to relationship extension—SAP streamlined its processes, reduced the sales cycle dramatically, and unlocked a lucrative SME market previously considered uneconomical.
SAP’s experience is a practical lesson in the transformative power of AI. Not only does this case study illustrate how embedding AI in business processes can unlock new market opportunities and drive efficiency, but it also warns of the consequences of falling behind. Companies that successfully integrate AI stand to gain significant competitive advantages, while those that don’t may struggle with longer sales cycles and diminished market relevance. For internal auditors, this is a dual opportunity to explore in your audit approach: ensuring that technology investments are well-governed and aligned with strategic business objectives, and to identify areas where process improvements can drive overall business performance.
Read more on this case study via HBR
A Quiz to Measure How Smart Chatbots Are (and You Can Test Your Capabilities Against GenAI)
This interactive quiz challenges users to gauge how their intelligence compares with that of modern chatbots powered by generative AI. It offers a playful yet thought-provoking way to engage with the rapidly evolving capabilities of AI.
While this quiz is primarily just a fun way to test your knowledge and see how you fare against AI, it also underscores the rapid pace at which AI capabilities are evolving. For internal auditors, it serves as a lighthearted reminder that the landscape of technology is continuously shifting, and staying informed is key to managing emerging risks and opportunities.
Take the quiz ! Are You Smarter Than AI?
Closing
Thank you for reading my weekly digest on AI in Internal Audit. I hope these reflections spark further dialogue on how emerging technologies intersect with internal audit practices. Feel free to share your thoughts or share this with other via the link below!