Curious AI 42
Oliver Rochford
Evangelist @ Auguria | Technologist | Cyberfuturist | Startup Advisor | Former Gartner Analyst
Welcome to issue 42 of the Curious AI Newsletter, curated by Oliver Rochford , Cyber futurist and former Gartner Research Director, and synthesized and summarized using AI.
AI Tribe of the Week
Data Drifters
Today’s digital nomads, navigating life by following AI’s every whim, no matter how baffling or absurd. These modern-day wanderers have ditched their internal compass in favor of a smartphone, letting algorithms decide things that would normally require basic common sense, like choosing pancakes or tofu for breakfast or moving to the middle of nowhere because "the data predicted a killer sunset." Ask them about their life plans, and they’ll look at you, wide-eyed, and say, “I’m just waiting for the algorithm to update.” You’ll find them in bizarre locations, probably on the edge of a cliff in rural Idaho, because "data science says it’s the optimal spot for spiritual awakening (with Wi-Fi)."
Slogan: "The algorithm knows best, or so I hope."
Want to discuss AI, quantum, and other emerging technologies?
Join our Slack
Most Ironic: AI's Quest to Save Humanity Might Destroy the Planet
As AI promises to solve humanity's biggest challenges, from disease to climate change, it's also threatening to accelerate environmental degradation. Data centers powering AI are set to triple their CO₂ emissions by 2030, spewing out 2.5 billion tons annually. Companies like Oracle are contemplating nuclear-powered data centers, while others scramble for renewable solutions. It's the ultimate irony: in our race to create machines smarter than ourselves, we might be outsmarting our chances for a sustainable future.
Key Takeaway:
The rapid expansion of AI technology is contributing to a significant increase in carbon emissions, highlighting a critical paradox: while AI aims to solve global challenges, it may be exacerbating environmental issues. It's essential for the tech industry to prioritize sustainable practices to ensure that the pursuit of artificial intelligence doesn't come at the expense of our planet's health.
Articles:
Most Overhyped: AI Startups Valued in Billions Without a Product
Welcome to the new gold rush, where an AI startup like Poolside can nab a $3 billion valuation without a single product. Traditional VCs are getting sidelined as tech giants throw cash at anything with "AI" in the name. Yet, enterprises are already hitting the brakes, abandoning 30% of generative AI projects by 2025 due to sky-high costs and nebulous benefits. We're in the brute force era of AI, throwing more GPUs at problems without clear solutions. Despite the promise of the tech, the business side may be a bubble waiting to burst, and if it does, the fallout will be spectacular.
Key Takeaway:
The astronomical valuations of AI startups without tangible products highlight a speculative bubble driven by hype rather than real-world value. This trend underscores the need for cautious investment and realistic expectations, as sustainable success in AI will depend on delivering proven solutions rather than capitalizing on excitement alone.
Articles:
Most Contradictory: AI Can't Summarize Documents but Can Debunk Conspiracies
Apparently AI can't accurately summarize a government document but can convince someone that the Earth isn't flat? Fascinating. I use AI weekly to summarize data, btw. But a recent Australian government trial found AI falls short in basic summarization tasks. Yet another study says chatbots can reduce belief in conspiracy theories by 20%. AI seems better at changing minds than processing information accurately. Perhaps we should recalibrate our expectations. Or just enjoy the irony 🤣.
Key Takeaway:
AI's capabilities are inconsistent and context-dependent; while it struggles with tasks requiring precise information processing, it can excel in influencing human beliefs and attitudes. This highlights the need to understand AI's strengths and limitations, leveraging it appropriately while remaining cautious of its shortcomings.
Articles:
The Curious AI now comes with data! I’ve begun building a trend analysis app to model the future horizons for AI. The future of AI development can be roughly categorized into four key paths based on current signals and trends:
Each of these paths represents a different expectation about where AI is heading, from pessimistic (plateaued) to optimistic (accelerating). Depending on which trends gain traction, AI will shape industries, governance, and daily life differently. Looking at current trends, it's reasonable to predict that we may not settle on just one of these paths but could experience a blend of gradual evolution with hype cycles mixed in.
AI is impacting different fields in different ways and at a different pace.
We see this already in the data, where different fields are experiencing the impact of AI in different ways.
In the chart above, for example, we can see that the greatest impact so far is being seen in the fields of AI warbots, AI and robotics, AI in the datacenter, AI Carbon Footprint, software development, and AI Business. This is not really surprising. We expect a new industrial revolution to begin in the supply chain, and indeed we see a high level of acceleration in datacenter buildout with a correlation with AI’s environmental impact, and demands on energy supply.
Sentiment: The good and the bad
What is also interesting to see is the sentiment in different domains.
We can readily see that fields with a lot of negative reporting are AI Business, AI in the Datacenter, AI Trust, Risk and Security Management, AI’s Carbon Footprint, AI Ethics, AI in Education, and AI warbots. This also makes sense. The reports and studies regarding the impact on education are indeed very mixed, with a lot of concern about how using technological crutches can negatively drive learning behavior, as well as worries about cheating and plagiarism. The energy use of AI is also a growing area of concern, with many reports indicating that GenAI has increased datacenter CO2 output by 200%-300%.
But we also see a lot of optimism and growing acceptance. See, for example, AI in healthcare, where neutral and positive sentiment articles dominate, and we have indeed seen positive progress. AI and robotics is another bright spot, and possibly a field experiencing a current wave of real breakthroughs.
Future signals point towards rapid progress.
When we look at the reporting in general, we see that all signals indicate progress, with a slight weighting towards “AI is accelerating.”.
We see that most articles indicate that AI is progressing, and most reports are neutral or positive in outlook. Optimistic views dominate on AI acceleration, which should also come as no surprise, as folks in that mindset tend to be the optimistic types. But what sticks out to me is the large neutral sentiment around AI evolving gradually. Possibly somewhat counterintuitively, that seems a stronger signal to me, as it’s based on more balanced research and data points.
As we add more articles, these signals should become stronger and easier to differentiate.
Learn more about the latest Quantum Technology
Check out the Intriguing Quantum Newsletter.
AI in Government
Nevada to Utilize Google AI for Unemployment Appeals Decisions
Nevada plans to implement Google's AI to analyze and recommend decisions for unemployment appeals, aiming to speed up the process and clear the backlog post-pandemic. Concerns of potential bias and accuracy challenges arise amidst hopes of faster claim resolutions.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
AI Copyright, Regulation, and Antitrust
US Proposes New AI Development and Cybersecurity Reporting Rules
The US Department of Commerce is proposing new rules requiring reporting on advanced AI models and computing clusters for national security and innovation enhancement. Stakeholders can provide feedback until October 11, 2024.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
AI Business
AI Giants Fuel Unprecedented Disruption in Venture Capital Landscape
Tech giants like Microsoft and Amazon are reshaping the VC market by heavily investing in AI startups, leading to a tough spot for traditional VCs amidst a slump and IPO drought.
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will accelerate
AI Bill Management: Preventing Cost Overruns and Addressing AI Ethical Concerns
Organizations adopting AI need to manage the financial and emotional costs while capturing productivity benefits, as advised by Gartner analysts during their Symposium event. Preventing cost overruns and addressing ethical concerns in AI operations are emphasized.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will evolve gradually
GenAI hype meets harsh reality as enterprises wrestle with business case
Enterprises struggle with the business case for generative AI projects as Equinix and Nvidia capitalize. Gartner predicts 30% of projects will be abandoned by 2025 due to high costs and unclear benefits. Organizations wary of AI's impact on IT budgets and investments raise concerns about project sustainability.
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI is hype
We're in the brute force phase of AI – once it ends, demand for GPUs will too
Gartner's research indicates that specialized AI hardware is doomed in the brute force phase, affecting GPU demand. Generative AI is limited to 5% of tasks, prompting a shift towards composite AI. The article discusses caution in using generative AI, favoring established AI approaches.
Sentiment: Critical | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
Agentic AI: A deep dive into the future of automation
Agentic AI, a fusion of generative AI with classical automation, is set to revolutionize enterprise operations. It promises a paradigm shift by enabling autonomous decision-making, creativity, and real-world actions.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
Releases and Announcements
ServiceNow Enhances Now Assist AI Platform with New Features
ServiceNow enhances Now Assist AI platform with customizable AI agents for enterprise workflows, providing control over AI-human collaboration in managing tasks. New features include data visualization, chat/email reply generation, and connection to Microsoft's Copilot for Microsoft 365.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
OpenAI's "Strawberry": A New Era in AI Reasoning
OpenAI's newly released AI model, o1-preview, known as "Strawberry," enables advanced reasoning capabilities for solving complex problems, surpassing human PhD experts in physics challenges. Despite limitations, it showcases significant progress in problem-solving tasks that require planning and iteration.
Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
AI and Society
Chatbots can persuade people to stop believing in conspiracy theories
AI chatbots can reduce belief in conspiracy theories by about 20%. MIT Sloan and Cornell University researchers used large language models to engage participants in discussions, resulting in a significant decrease in belief in baseless theories.
Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
Recommended by LinkedIn
AI Carbon Footprint
Datacenters Facing 3x CO2 Emissions Increase Due to Generative AI
The datacenter industry faces a 3x increase in CO2 emissions due to generative AI development, leading to 2.5 billion tonnes of emissions by 2030. Morgan Stanley estimates a market for carbon capture solutions valued at $45B, highlighting the impact on global emissions and opportunities for decarbonization.
Sentiment: Negative | Time to Impact: Mid-term | Future Signal: AI will accelerate
Elon Musk's xAI Supercomputer Sparks Controversy in Memphis
Elon Musk's xAI supercomputer project in Memphis faces controversy due to pollution concerns. Musk's secretive approach and lack of community involvement have raised environmental worries, with infrastructure challenges and potential impacts on marginalized neighborhoods.
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will accelerate
Generative AI is reportedly tripling carbon dioxide emissions from data centers
Generative AI's expansion triples data centers' CO2 emissions, foreseeing 2.5B tons by 2030. Challenges arise in meeting net-zero goals as energy-demand soars. Future AI tech and carbon removal investments crucial for environmental targets.
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will accelerate
Oracle is designing a data center that would be powered by three small nuclear reactors
Oracle plans to build a data center powered by three small nuclear reactors, driven by increasing electricity demand from AI. Small modular reactors promise reliable, carbon-free power, but face commercialization challenges.
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will accelerate
DOE Roadmap for AI's Energy Impact on Data Centers and Sustainability
The US Department of Energy (DOE) collaborates with industry experts to tackle AI's surging energy demands on data centers, emphasizing sustainability and grid reliability. The report details strategies to address escalating power needs driven by AI applications like large language models, proposing partnerships for efficient energy algorithms and clean energy
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
AI Copyright, Regulation, and Antitrust
Senate Urges FTC to Investigate AI Content Summaries for Anticompetitive Practices
The Senate is pressuring the FTC and Justice Department to look into AI-generated content summaries like news and recipes, citing potential anticompetitive implications on creators and publishers due to major platforms' dominance. They question if these practices violate antitrust laws.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
AI Ethics
White House extracts voluntary commitments from AI vendors to combat deepfake nudes
The White House obtains commitments from major AI vendors to combat deepfake nudes and child sexual abuse material, aiming to source datasets responsibly and remove inappropriate content from AI training datasets.
Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will evolve gradually
AI Healthcare
AI Technology Revolutionizes Disease Detection: Over 1,000 Conditions Predicted Early
New AI technology, MILTON, detects early signs of 1,000+ diseases using patient data, aiding in more effective treatments. AstraZeneca's tool predicts diseases like COPD and dementia years in advance, enhancing preventive care.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
Lumin Fitness Launches Innovative AI Coaches to Transform Your Workout Experience
Lumin Fitness introduces virtual AI coaches revolutionizing workout experiences with tailored guidance and feedback via LED screens. The system incorporates gamified elements, personalized training, and real-time form monitoring. Members can choose coaches based on voice and style preferences through a mobile app.
Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate
AI Trust, Risk, and Security Management
MI6 and CIA Utilize Generative AI Against Tech Threat Actors
MI6 and CIA employ generative AI to enhance intelligence activities against tech-driven threat actors. They address challenges from Russia and China using AI for data analysis and protecting operations. The article highlights their joint efforts to combat evolving threats in the digital age.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI and Robotics
Teaching Robots Efficiency: The Rise of "Lazy Robotics"
Lazy robotics, championed by René van de Molengraft, focuses on making robots more efficient by teaching them to prioritize essential data over non-essential. This approach, exemplified through robots playing soccer, aims to enhance performance, decrease energy usage, and improve reliability.
Sentiment: Positive | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
AI at Work and Employment
AI generating a ‘sea of sameness’ in job applications
New technologies like generative AI are streamlining job application creation but leading to a "sea of sameness" in content. AI tools offer both benefits and challenges, requiring users to find a balance in standing out while avoiding gimmicks.
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI in Education
AI-Generated Essays: Unleashing a Solution We're Overlooking
AI-generated essays, notably from ChatGPT, pose a challenge in academia. OpenAI's watermarking solution could address plagiarism by making AI text distinguishable. However, reluctance to release the technology due to competition hampers progress.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will evolve gradually
Teacher's Debate on Students Using ChatGPT for Self-Introductions
Prof. Fritts caught students using ChatGPT for simple intros in an ethics class, igniting debate on AI's role in education. Some educators worry about AI impacting critical thinking skills.
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI in Finance
Banks Are Embracing AI. Here’s What You Need to Know.
Banks face challenges with AI implementation, impacting customer experience. AI in banking offers operational efficiency and fraud detection but poses regulatory and chatbot challenges. Responsible AI integration is crucial for success.
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
AI in Software Development
AI Startup Poolside Approaches $3 Billion Valuation Before Product Launch
AI startup Poolside, co-founded by former GitHub CTO Jason Warner and software entrepreneur Eiso Kant, is set to raise $500 million, valuing the company at $3 billion before releasing a product. The Paris-based company aims to develop advanced AI for software coding.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
AI in the Data Center
The Battle Begins For AI Inference Compute In The Datacenter
Cloud builders and hyperscalers are gravitating towards AI inference due to high costs. Companies like Groq, Cerebras, and SambaNova are emerging with advanced hardware for more efficient inference. The competition focuses on performance and cost-effectiveness, crucial for widespread AI adoption and profitability in the future.
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
Opportunity Arises in AI Data Centers' Power Revolution
Entrepreneurs and investors see a major opportunity in the power-hungry AI data centers as startup Exowatt unveils a solar storage solution. Big Tech's choices between traditional grid power and clean energy options impact billions, with AI data centers potentially boosting carbon capture tech usage.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
Interesting Papers & Articles on Applied AI
Reinforcement Learning in Data Centers: Optimizing Energy Efficiency
Reinforcement learning aids Meta in optimizing data center environmental controls, reducing energy consumption and water usage. The approach improves cooling efficiency, crucial for sustainability. Simulator-based reinforcement learning achieved 20% fan energy savings and 4% water usage reduction.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
AI worse than humans in every way at summarising information, government trial finds
The Australian government trial reveals that AI is inferior to humans in summarising documents, potentially creating more work for humans. The study compared AI and human summaries on various criteria, with humans outperforming in every aspect. The findings emphasize the need for AI to augment, not replace, human tasks.
Sentiment: Critical | Time to Impact: Short-term | Future Signal: AI is hype
Can Large Language Models Generate Novel Research Ideas?
Large language models (LLMs) are evaluated on their capability to generate novel research ideas compared to human experts. LLMs produced more novel ideas but slightly less feasible. Challenges in LLM self-evaluation and diversity exist.
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.