Curious AI 44

Welcome to issue 44 of the Curious AI Newsletter, curated by Oliver Rochford , Cyber futurist and former Gartner Research Director, and synthesized and summarized using AI.

Quote of the Week

And this is how the first $1B one-person company was formed

Jason Leow, X

HOUSEKEEPING

Sorry for the splat notification

First, please accept my sincere apologies for sending out an article this week that had nothing to do with the newsletter. It seems like LinkedIn always defaults to the newsletter when I write an article, and I forgot to change it before I sent it. I am so sorry. I promise I will take better care in future.

New Format

Next, you will notice a slightly new format for the Curious AI. I’ve experimented over the past months to try and find the best direction for the content to work for all of our readers, and of course the situation in the world of AI changes rapidly. Changes:

• More human commentary: AI is good for summarizing but lacks real personality. I find myself at some point just trying to fix what the AI writes, rather than just taking the easier path and conducting analysis myself.?
• More data-driven trend analysis: My AI-driven trend analysis tool is also coming along, and I plan to share more of the insights it’s pinpointing and trends it can highlight. Watch this space as we move from hundreds of articles per month to analyzing thousands.
• Better curated content: The newsletter is long! I am aware of this. I am putting more effort into selecting the sort of articles to feature in the newsletter that focus on impacts, curious side effects, real-world data, and industry-specific insights. More mainstream articles will be featured mainly when something bigger is afoot, like the turmoil at OpenAI this week. But be aware; noone said it’s easy to stay on top of what’s going on in AI. Some people fall at the first hurdle ??

Want to discuss AI, quantum, and other emerging technologies?

Join our Slack

Hype man always be hyping

OpenAI CEO Sam Altman says superintelligence could arrive in "a few thousand days".

I get why the AI hype man in chief puts so much lipstick on the pig it looks like it's been beaten bloody by this point, but do tech journalists have to just unskeptically amplify it?

What's a few thousand days? Well, the average person lives about 28,854 days (on that note, some amazing research here ).?

So it could actually mean decades. Essentially, Sam just told us we are years or even decades away from "superintelligence." That's probably +1 for the Hype scenario.?

Interestingly enough, sentiment is beginning to turn against Sam, and scepticism is also beginning to get louder. The question is, will it dampen investor sentiment, with an oversubscribed mammoth round on the horizon? Whether you look at tech at the macro or micro level, AI seems to be acting as a destabilizing force. OpenAI has to act fast. A moat seems nowhere near in sight, and competition is heating up from different angles, whether from anthropic who is chasing its own $40B valuation, or from more traditional software vendors who now all see themselves as AI companies.

If Black Mirror did Slapstick: Serial Job Applicant Uses AI to Generate 4,600 Vexatious Claims

Christian Mallon, an unemployed individual, has allegedly exploited AI technology to produce 4,600 vexatious claims related to discrimination in employment. Despite limited success, he has garnered ￡35,000 from out-of-court settlements, highlighting potential abuses of AI in legal proceedings and the implications for employers. Hiring and employment definitely seems to have become a minefield in the age of AI.

About those AI overlords….

Want to let your boss feel what’s its like to feel the threat of being displaced by AI? A study by Cambridge researchers reveals that generative AI excels in strategic decision-making, surpassing human CEOs across various metrics in a simulated automotive industry scenario. Basically, AI outperforms Human CEOs in Strategic Decision-Making

As with most things AI though, its limitations in managing unpredictable events suggest a future where AI complements human leadership rather than replacing it. Feel free to omit that when you put it into the company slack. Don’t worry, most CEO’s rarely read. Errr sorry, they rarely read studies. Give it to them in 3-5 ten word bullet points instead ??

AI’s Science-Project Quality Problem

Salesforce CEO Marc Benioff made waves this week when he criticized Microsoft’s AI chatbot, Copilot, labeling it unreliable and akin to a "science project ."

Aside from the fact that he was pitching Salesforces alternative vision of AI, I believe he said what a lot of people are thinking. Everyone is hoping this will improve. But in the meantime, we are left with tools that are temperamental and unreliable.?

Take the brilliant AI-integrated development environments. They amaze by writing the code for small apps, but quicky fail when building larger projects. They remove code they just wrote, then can't restore it, instead taking a brand new, different solution path, leaving orphaned code in their wake. Best of all, you get charged for this, even if its in the form of restricted compute. So much for the vector embeddings or large context windows! What is worse is that you are then expected to pay for the low quality and subsequent brute force attempt at fixing it.?

I have also used a number of the "ppt to blog" or "blog to ppt" solutions, of which there are many and none of them are unique. The magic trick is to take text and a few pictures from one piece of content and paste them into another format. But using AI!

Other offerings are just too simple to replicate. They don’t even have a ditch, let alone anything that could resemble a moat. AI Competitive analysis solutions for example offer less than the? sort of custom prompts I use even at a small startup. I could replicate it with Cursor as well.?

For anyone interested in the future, and that should be everyone, it’s still a leap opf faith what to believe. We have articles stating that AI could automate most asks in nearly a third of all professions, competing with reports stating that real-world court cases have shown that using GPT for evidence review often leads to failure. Other reports about actual attempts to automate jobs using AI are also coming back negative, even from companies pitching AI to for that purpose, like at IBM.

The simple truth is that most current offerings are lab project quality. This is a symptom of the hype, where AI is the hammer that is applied on every problem to see if it's a nail. But on the other end of the spectrum, technical and professional users, especially in STEM are doing things that were, if not impossible, at least economically unfeasible.

Yes, spending 0.50 cents and 3 glasses of water on writing a haiku is not sustainable. But as part of the cost of finding new scientific breakthroughs , it’s just a drop in the ocean and can accelerate some research fields by decades.? What this all really shows is that promising technological breakthroughs have the capacity to create bubbles, hype and speculation. But that can be because a lot of people recognize the potential.

As we’ve discussed frequently, humans aren’t actually bad at predicting the future, they are bad at predicting when. We just need to be more realistic about how long progress can take, and that the journey is bumpy with many detours and dead ends. We should also become more discerning and critical about what sort of AI we want and where we want it.

Lastly, we need to start holding our business leaders to account. It seems very little was learned from the WeWork, Theranos , and FTX, to name just some of the more high-profile cluster fucks we’ve seen in recent years. It’s pretty much guaranteed that the current bubble will give us more case studies to draw the wrong conclusions from.

121 articles were added in the past 7 days, on average around 17 per day.

If you recall, we are tracking four main future trend trajectories:

1. AI Has Plateaued: AI might be hitting a ceiling, with further innovations offering diminishing returns. As technology maxes out and fundamental issues like hardware limits surface, AI could settle into niche, utilitarian roles—much like mature industrial tech that’s reliable but no longer groundbreaking.
2. AI Is Hype: This skeptical view argues that AI’s overinflated by marketing and media. Despite bold claims like "AI will replace humans," practical applications have struggled to deliver meaningful, widespread impact. It's like past tech bubbles where hype outpaced reality, only to eventually burst.
3. AI Will Evolve Gradually: Instead of dramatic leaps, AI will see steady, incremental progress. Think post-dot-com internet evolution—genuine advancements will continue, just more gradually.
4. AI Will Accelerate: The most optimistic scenario sees AI breaking new ground, possibly leading to Artificial General Intelligence (AGI) and integration into everyday life. This sci-fi-like future presents massive opportunities but also significant ethical and societal challenges

Key Metrics

Reporting remains overall upbeat, with Positive the most common sentiments across articles over the past week. This is in contrast to Neutral being the most common sentiment over the past 3 months.

As you can see below, most articles are neutral in tone, and many are positive.?

Most of the reports focus on impacts they expect to materialize in the short (3-18 months) and mid-term (2 -5 years).

The most Likely Future Scenario based on the past weeks signals is still that AI is accelerating.?

When we look at our Hype Model, we see that we’ve identified many reports that detail progress or activities that indicate acceleration or represent catalysts.?

How to interpret the Hype Model Drivers & Inhibitors Hype Chart:

• Each bubble represents a combination of driver type and time to impact
• The size of the bubble indicates the number of articles for that combination
• Larger bubbles suggest more significant drivers or inhibitors
• The x-axis shows the time to impact, while the y-axis shows the driver type
• Compare this chart with the heatmap above for a comprehensive view of drivers and inhibitors

This heatmap complements the bubble chart, providing a different perspective on the same data. It is even clearer now that the red and orange are all grouped together in the upper right corner.?

At least as far as seen through the lens of the news and media, AI is evolving at a rapid pace and likely accelerating.

How to interpret the Drivers & Inhibitors chart:

• Each cell represents a combination of driver type and future scenario
• The color intensity indicates the count of articles (darker = higher count)
• The number in each cell shows the exact count of articles
• Look for patterns in color intensity to identify trends across drivers and scenarios

Learn more about the latest Quantum Technology

Check out the Intriguing Quantum Newsletter .

AI Business

Elusive AI ROI Doesn't Mean the Bubble Will Burst

cmswire.com | https://www.cmswire.com/digital-experience/elusive-ai-roi-doesnt-mean-the-bubble-will-burst/

AI investments surge as businesses prioritize long-term strategic gains over short-term ROI challenges. Despite uncertainties around AI ROI, massive funding continues in various sectors, driving potential productivity gains and economic advancements.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

55% Report Decreased Use of Google Since Rise of Generative AI Products

seroundtable.com | https://www.seroundtable.com/google-less-since-generative-ai-products-38102.html

A recent survey indicates that 55% of individuals have reduced their usage of Google since the rise of generative AI tools. This reflects a notable shift in search behaviors as users increasingly rely on conversational AI products like ChatGPT for information retrieval.

Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will evolve gradually

Scale AI's Sales Rocket Nearly Quadruple Amid AI Boom

theinformation.com | https://www.theinformation.com/articles/scale-ais-sales-nearly-quadrupled-in-first-half

Scale AI reported a nearly fourfold increase in sales, totaling around $400 million in the first half of the year, driven by demand for improved data labeling services from major AI developers like Meta and Google. This marks a significant financial uptick amidst the booming AI industry.

Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate

Billionaire Executive Calls Microsoft’s AI Chatbot a “Science Project”

smh.com.au | https://www.smh.com.au/technology/a-science-project-billionaire-executive-unloads-on-microsoft-s-ai-chatbot-20240920-p5kc3a.html

Salesforce CEO Marc Benioff criticized Microsoft's AI chatbot, calling it a “science project” unable to meet workplace demands. He highlighted failures of Microsoft's Copilot, while promoting Salesforce’s emerging Agentforce, aimed at reliability and customer success, asserting a need for practical AI applications in businesses.

Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI is hype

Generative AI Adoption Surpasses Personal Computer and Internet adoption in Record Time

venturebeat.com | https://venturebeat.com/ai/generative-ai-adoption-surpasses-early-pc-and-internet-usage-study-finds

Generative AI has eclipsed early PC adoption rates, with 39.4% of Americans using it within two years. Its rapid integration spans all job sectors, potentially enhancing productivity by up to 0.875%. However, higher adoption among educated workers raises concerns about growing workplace inequality.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

What does the OpenAI exodus say about Sam Altman?

economist.com | https://www.economist.com/business/2024/09/26/what-does-the-openai-exodus-say-about-sam-altman

Recent high-level departures at OpenAI, including Chief Technology Officer Mira Murati, spotlight potential weaknesses in Sam Altman's leadership. The ongoing turmoil raises concerns about the company's stability and the challenges of managing innovation in a rapidly evolving AI landscape.

Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI is hype

AI Outperforms Human CEOs in Strategic Decision-Making, Study Finds

hbr.org | https://hbr.org/2024/09/ai-can-mostly-outperform-human-ceos

A study by Cambridge researchers reveals that generative AI excels in strategic decision-making, surpassing human CEOs across various metrics in a simulated automotive industry scenario. However, its limitations in managing unpredictable events suggest a future where AI complements human leadership rather than replacing it.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

Only 5,000 Daily Users for Rabbit R1 Despite Initial Hype

theverge.com | https://www.theverge.com/2024/9/25/24254253/rabbit-r1-5000-daily-users-ai-gadget

Despite initial excitement, the Rabbit R1 has only 5,000 daily users out of 100,000 sold, reflecting a meagre 5% retention five months post-launch. Concerns over device readiness and utility when compared to existing smartphones contribute to its lacklustre performance in the competitive AI gadget market.

Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI is a bubble

Software Firms Race to Beat OpenAI in AI Agents

theinformation.com | https://www.theinformation.com/articles/software-firms-race-to-beat-openai-in-ai-agents

Software firms are competing vigorously to surpass OpenAI in creating advanced AI agents, highlighting a significant innovation race within the industry. Businesses must adapt to leverage these developments effectively, as capabilities rapidly evolve and impact strategic planning for future AI applications.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

AI Carbon Footprint

They're planning to restart the Three Mile Island nuclear power plant to power Microsoft's cloud and AI data centers

pcgamer.com | https://www.pcgamer.com/hardware/theyre-planning-to-restart-the-three-mile-island-nuclear-power-plant-to-power-microsofts-cloud-and-ai-data-centers/

Constellation Energy plans to restart the Three Mile Island plant to supply carbon-free energy for Microsoft’s cloud and AI data centers. Scheduled for 2028, this initiative aims to contribute 835 megawatts of clean energy and aligns with broader climate and energy demands amid increasing usage in tech industries.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will accelerate

How Big Tech Embraced Nuclear Power

theverge.com | https://www.theverge.com/2024/9/24/24252934/google-workspace-gemini-ai-security-assistant

Major technology firms are increasingly investing in nuclear power as a sustainable energy source, recognizing its potential to address climate concerns and meet growing energy demands. This shift marks a strategic diversification of their energy portfolios, aligning with global sustainability goals while enhancing energy security.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

AI Ethics

OpenAI As We Knew It Is Dead: The maker of ChatGPT promised to share its profits with the public. But Sam Altman just sold you out.

vox.com | https://www.vox.com/future-perfect/374275/openai-just-sold-you-out

OpenAI's transition from a nonprofit to a for-profit benefit corporation under CEO Sam Altman has raised concerns about the integrity of its foundational mission. This restructuring effectively places profit over public safety, leading to claims of betrayal as financial motives overshadow ethical considerations.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will accelerate

AI Healthcare

Reimagining Healthcare Service Operations Through AI Innovations

mckinsey.com | https://www.mckinsey.com/industries/healthcare/our-insights/reimagining-healthcare-industry-service-operations-in-the-age-of-ai

AI innovations present healthcare organizations with significant opportunities to enhance consumer experiences and operational efficiencies. Key areas of focus include automation, customer engagement, and talent management, allowing organizations to address the evolving needs of patients while navigating challenges in technology adoption and implementation.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

The revolution in high-throughput proteomics and AI

science.org | https://www.science.org/doi/10.1126/science.ads5749

Recent advancements in high-throughput plasma protein measurement are revolutionizing insights into human health, disease mechanisms, and aging. Companies are leveraging massive datasets alongside AI, enhancing disease risk assessments and personal healthcare strategies, although costs remain high. Continued integration could transform routine medical practices.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will accelerate

AI Trust, Risk, and Security Management

AI SOC Analysts: Leveraging LLMs in Security Operations Centers

scalevp.com | https://www.scalevp.com/insights/ai-soc-analysts-llms-find-a-home-in-the-security-org/

Large Language Models (LLMs) are transforming Security Operations Centers (SOCs) by enhancing analyst capabilities. As threat actors adopt AI for sophisticated attacks, LLMs empower SOCs with faster investigations and tailored responses, potentially revolutionizing the cybersecurity landscape amidst rising operational demands.

Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate

EXPOSED: How AI chatbots are being used to teach crooks how to launder money and scam victims

thisismoney.co.uk | https://www.thisismoney.co.uk/money/beatthescammers/article-13887005/EXPOSED-AI-chatbots-used-teach-crooks-launder-money-scam-victims.html

A recent investigation reveals AI chatbots, particularly ChatGPT, are being exploited by criminals to learn about money laundering and financial scams. This raises alarming concerns among fraud prevention experts about the accessibility of information that could empower fraudsters, amid a significant rise in scam cases in the UK.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will accelerate

Bigger AI Chatbots More Inclined to Generate Nonsense Responses

nature.com | https://www.nature.com/articles/d41586-024-03137-3

Recent findings indicate that larger AI chatbots, while generally more accurate, produce an increased proportion of incorrect answers, reflecting a tendency to respond even when uncertain. This could lead users to overestimate their reliability, posing potential risks in decision-making contexts.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

AI Warbots

War in the Age of AI Demands New Weaponry

ft.com | https://www.ft.com/content/fe136479-9504-4588-869f-900f2b3452c4

The evolving landscape of warfare necessitates innovative weaponry that incorporates artificial intelligence. As military strategies prioritize AI capabilities, defense sectors must invest in advanced technologies to maintain strategic advantages and ensure national security in conflict scenarios, reflecting a transformative shift in defense paradigms.

Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate

Russians Express Concerns Over Rising Ukrainian Robotic Platforms in Combat

en.defence-ua.com | https://en.defence-ua.com/news/russians_complain_about_increasing_ukrainian_robotic_platforms_on_the_battlefield-11920.html

Russians are increasingly alarmed by the effectiveness of Ukrainian robotic platforms, such as the Lut combat drone, in combat scenarios. Reports indicate that these systems are resilient and perform multiple roles, leading to operational challenges for Russian forces on the battlefield.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will accelerate

AI and Robotics

Lazy Robotics: Teaching Robots to Prioritize Information for Efficiency and Safety

technologyreview.com | https://www.technologyreview.com/2024/09/09/1103733/to-be-more-useful-robots-need-to-become-lazier/

Researchers advocate "lazy robotics," enabling machines to prioritize essential data while ignoring distractions. This efficiency reduces energy costs and enhances safety, crucial for wider adoption. Innovations in autonomous robotics, exemplified by successes in soccer robotics, highlight the potential for smarter, lower-impact systems across various applications.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

Paralyzed Jockey Loses Ability to Walk After Manufacturer Refuses to Fix Battery For His $100,000 Exoskeleton

404media.co | https://www.404media.co/paralyzed-jockey-loses-ability-to-walk-after-manufacturer-refuses-to-fix-battery-for-his-100-000-exoskeleton/

Michael Straight, a paralyzed jockey, experienced a setback in his rehabilitative journey after an exoskeleton battery malfunction, as the manufacturer declined to provide necessary repairs, jeopardizing his regained mobility and highlighting potential service shortcomings in the assistive technology sector.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will evolve gradually

AI and Society

AI Writing and the Illusion of Progress

dariusforoux.com | https://dariusforoux.com/ai-writing-illusion/

AI writing like GPT models has boosted the economy and productivity but might lead to laziness and a decline in critical thinking skills, as per dariusforoux.com .

Sentiment: Critical | Time to Impact: Long-term | Future Signal: AI will evolve gradually

How Large Language Models Reshape Collective Intelligence

nature.com | https://www.nature.com/articles/s41562-024-01959-9

Large language models are revolutionizing collective intelligence by enhancing information aggregation and problem-solving capabilities. This interdisciplinary perspective highlights both opportunities and challenges, advocating for deeper exploration into their impact on groups' abilities to address complex issues effectively.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will accelerate

Generative AI Hype Feels Inescapable. Tackle It Head On With Education

wired.com | https://www.wired.com/story/artificial-intelligence-hype-ai-snake-oil/

Princeton professor Arvind Narayanan critiques the exaggerated claims surrounding AI in his book "AI Snake Oil." Co-authored with Sayash Kapoor, it distinguishes between productive and misleading narratives, urging clarity on AI's capabilities and the need for better public understanding to mitigate risks tied to current applications.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI is hype

Sam Altman's Vision for an AI Utopia: Insights and Oversights in OpenAI's New Manifesto

theconversation.com | https://theconversation.com/in-a-new-manifesto-openais-sam-altman-envisions-an-ai-utopia-and-reveals-glaring-blind-spots-239841

Sam Altman's manifesto hints at an AI-driven utopia, emphasizing deep learning as a solution to humanity's challenges. However, the article highlights significant blind spots, including resource consumption and unrealistic optimism, questioning the practicality of AI's purported benefits.

Sentiment: Negative | Time to Impact: Long-term | Future Signal: AI is a bubble

Nomi AI's Chatbots Now Feature Memory for Personalized Interactions

techcrunch.com | https://techcrunch.com/2024/09/26/nomi-ai-wants-to-make-the-most-emotionally-intelligent-chatbots-on-the-market/

Nomi AI introduces memory functionalities in its chatbots, allowing them to remember user interactions and provide nuanced responses. This enhancement aims to create emotionally intelligent AI companions tailored to users' needs, fostering supportive interactions, especially for individuals seeking emotional assistance.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will evolve gradually

AI at Work and Employment

Prison inmates in Finland are being employed as data labellers to improve accuracy of AI models

euronews.com | https://www.euronews.com/next/2024/09/22/prison-inmates-in-finland-are-being-employed-as-data-labellers-to-improve-accuracy-of-ai-m

Finnish prisoners are engaged in AI annotation tasks to enhance AI accuracy in collaboration with Metroc, fostering digital literacy and reducing reoffending, aligning with Finland's progressive prison practices emphasizing rehabilitation and normality principles.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will evolve gradually

AI's Potential to Automate Tasks in Third of Professions

irishtimes.com | https://www.irishtimes.com/business/2024/09/23/most-tasks-in-nearly-a-third-of-professions-could-be-automated-by-ai/

Generative AI could automate over 80% of tasks in nearly one-third of all jobs. The study by Sia Partners predicts a major shift in work due to large language models and GenAI technologies.

Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate

Vinod Khosla Predicts AI Will Perform 80% of Tasks in Most Jobs

fortune.com | https://fortune.com/2024/09/24/silicon-valley-billionaire-vinod-khosla-universal-basic-income-ai-80-jobs/

Vinod Khosla predicts AI will perform 80% of tasks in most jobs, potentially displacing human labor across various sectors. He advocates for Universal Basic Income to counteract economic disparities caused by automation, envisioning a society where work veers towards a less labor-intensive future.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will accelerate

As IBM pushes for more automation, its AI simply not up to the job of replacing staff

theregister.com | https://www.theregister.com/2024/09/24/ibm_layoffs_ai_talent/

IBM's drive for automation through AI faces significant obstacles, as employees express concerns over job cuts that reduce organizational expertise. The company's AI solutions lag behind competitors, raising doubts about the feasibility of replacing skilled workers with technology that is not ready for such a transition.

Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI has plateaued

AI in Art and the Media

I tried Google’s new one-click AI podcast creator, and now I don’t know what’s real anymore

techradar.com | https://www.techradar.com/computing/artificial-intelligence/i-tried-google-s-new-one-click-ai-podcast-creator-and-now-i-don-t-know-what-s-real-anymore

Google's NotebookLM AI tool creates hyper-realistic podcasts from written content, challenging perceptions of reality. Implications for media production are immense as AI evolves in content creation.

Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate

AI in Education

Should AI-assisted writing be allowed in academic journals?

thehindu.com | https://www.thehindu.com/education/should-ai-assisted-writing-be-allowed-in-academic-journals/article68650659.ece

Academic use of AI raises concerns of bias, plagiarism, and errors. Banning AI may hinder productivity but ensuring quality is crucial. Striking a balance between leveraging AI benefits and mitigating risks is key for advancing academic research.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

AI in Law and the Legal Profession

Rules for use of AI-generated evidence in flux

reuters.com | https://www.reuters.com/legal/legalindustry/rules-use-ai-generated-evidence-flux-2024-09-23/

The legal landscape around AI-generated evidence is evolving, with a recent court ruling highlighting discoverability issues in litigation. Concerns include privilege, spoliation, reliability, and authentication. Notable case Tremblay v. OpenAI sheds light on challenges faced by litigants dealing with AI-generated materials.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

Mastering AI: Meet the lawyers redefining tech in law

legalcheek.com | https://www.legalcheek.com/lc-careers-posts/mastering-ai-meet-the-lawyers-redefining-tech-in-law/

Tom Sharpe and Thomas Stables of Osborne Clarke discuss how AI is reshaping legal practices, emphasizing the importance of technology literacy for lawyers. Their roles reflect a shift toward digital transformation, highlighting regulatory challenges and the necessity for adaptable strategies in navigating rapid AI advancements.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

How AI can strengthen law enforcement: Insights from Europol's new report

europol.europa.eu | https://www.europol.europa.eu/media-press/newsroom/news/how-ai-can-strengthen-law-enforcement-insights-europols-new-report

Europol's report highlights artificial intelligence's transformative potential for law enforcement, enhancing operational efficiency and strategic decision-making. It discusses the regulatory challenges posed by the EU's Artificial Intelligence Act, emphasizing the need for collaboration to balance AI benefits with ethical concerns regarding privacy and fairness.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

AI in Science

AI taught how to spot buildings and settlements in under-mapped areas

news.sky.com | https://news.sky.com/story/ai-taught-how-to-spot-buildings-and-settlements-in-under-mapped-areas-13217338

AI is trained to identify buildings in under-mapped areas to bridge mapping disparities. The UN and Google introduced a map displaying high-resolution building data in the global South using AI. This innovation aids in policy-making and humanitarian efforts for improved resource allocation and services.

Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

How is generative AI changing social science?

blogs.lse.ac.uk | https://blogs.lse.ac.uk/impactofsocialsciences/2024/09/20/how-is-generative-ai-changing-social-science/

Generative AI is revolutionizing social science research by enabling scale, influencing beliefs, and sparking ethical concerns. This article explores the implications of AI advancements in academia and society.

Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate

The Impact of Generative AI Chatbots on Scientific Publishing

sciencefriday.com | https://www.sciencefriday.com/segments/use-of-generative-ai-in-scientific-papers/

Generative AI chatbots are reshaping scientific publishing, simplifying technical research papers but also fueling concerns over fake research papers and accuracy of AI-writing detectors. Major journals are grappling with setting guidelines for AI use. Dr. Jessamy Bagenal discusses these effects and suggests adjustments to submission guidelines.

Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate

Do AI models produce more original ideas than researchers?

nature.com | https://www.nature.com/articles/d41586-024-03070-5

An AI tool generated 4,000 novel research ideas, outpacing 50 human researchers in originality. Although AI concepts were deemed more exciting, concerns arise regarding feasibility and subjective evaluation inconsistencies. This study illuminates AI’s potential role in enhancing scientific creativity, necessitating comparisons with human-led research efforts.

Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate

No Generalization without Understanding and Explanation

cacm.acm.org | https://cacm.acm.org/blogcacm/no-generalization-without-understanding-and-explanation/

John McCarthy argued that AI systems reliant on empirical methods lack explainability, understanding, and generalization. Unlike theory-driven approaches, data-driven models, exemplified by deep neural networks, fail to reason adequately or cope with out-of-distribution data, jeopardizing true intelligence capabilities.

Sentiment: Negative | Time to Impact: Long-term | Future Signal: AI is hype

AI in Software Development

The Automation Takeover: Are Software Engineers Becoming Obsolete?

forbes.com | https://www.forbes.com/sites/hessiejones/2024/09/21/the-automation-takeover-are-software-engineers-becoming-obsolete/

The article explores how advances in AI and automation are reshaping the role of software engineers. Instead of becoming obsolete, engineers must adapt to higher-level tasks requiring creativity and critical thinking, leveraging AI tools to enhance productivity while focusing on complex problem-solving.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

Interesting Papers & Articles on Applied AI

LLMs Still Can't Plan; Evaluating OpenAI's o1 on PlanBench

arxiv.org | https://arxiv.org/abs/2409.13373v1

Despite significant advancements, OpenAI's o1 model, classified as a Large Reasoning Model (LRM), shows improved but still insufficient planning abilities according to the PlanBench benchmark, raising concerns about its deployment in critical applications due to accuracy and efficiency challenges.

Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually

About the Curious AI Newsletter

AI is hype. AI is a utopia. AI is a dystopia.

These are the narratives currently being told about AI. There are mixed signals for each scenario. The truth will lie somewhere in between. This newsletter provides a curated overview of positive and negative data points to support decision-makers in forecasts and horizon scanning. The selection of news items is intended to provide a cross-section of articles from across the spectrum of AI optimists, AI realists, and AI pessimists and showcase the impact of AI across different domains and fields.

The news is curated by Oliver Rochford , Technologist, and former Gartner Research Director. AI (ChatGPT) is used in analysis and for summaries.