Curious AI #49
Oliver Rochford
Evangelist @ Auguria | Technologist | Cyberfuturist | Startup Advisor | Former Gartner Analyst
Welcome to the fourtyninth issue of the Curious AI Newsletter, curated by cyber futurist and former Gartner Research Director Oliver Rochford , and synthesized by AI.
Welcome To The Age of Megaloman.ai
News broke this week that Elon Musk is in talks with Middle East investors to secure funding aimed at positioning xAI's valuation at $45 billion. XAI already raised $6 billion in series B funding in May, reaching a post-money valuation of $24 billion. If you take away the “relationship” revenue that xAI has, including from Tesla and other Musk properties or allies, it’s hard to consider xAI a major AI player, remote-controlled robots aside. Perplexity, the search engine for analytically challenged people, has just managed to capture 100 million weekly searches but is only valued at roughly $8 billion, despite a grab for the potentially lucrative search advertising market. OpenAI, the LLM pioneer and? $150 billion behemoth, is also pushing into search, even as Sam Altman is warning that a lack of compute is already delaying product launches.?
So maybe Elon is betting on having an advantage there with his massive 100K GPU supercluster. There’s even talk of doubling the capacity, although it’s unclear where the additional GPU’s will come from or how the many other dependencies will be met. And what has xAI shown us since May? Well, there’s Grok, the pottymouthed “antiwoke ” chatbot that Musk is now recommending as a medical advisor (despite radiologists warning against relying on the result), and an API .? Did I mention remote-controlled “robots”??
Maybe he can get Masayohsi Son to invest. He is bullish on AI and believes that an “artificial superintelligence” will exist by 2035 ?? But then he also believed that WeWork wasn’t just a coworking company masquerading as a tech startup. Or maybe he did understand, and was hoping everyone else wouldn’t. Either way,? I think it is fair to say that at this stage we are just making these numbers up based on very optimistic assumptions about the pace of progress and future potential of AI. As we’ve mentioned before, it is entirely possible (and historically not unique) to experience a bubble even when a new technology is disruptive in potential. The Dotcom bubble is often seen as heralding the Internet? and digitization. Right now, we may be going through the age of Megaloman.ai .?
What is AI Art worth?
Another preposterous story ?this week is about Botto, an autonomous AI artist. A report from Sotheby's says that the AI Artist made $351,600 in sales. And Botto is not alone. Late this month, Sotheby's will auction off the first piece of art credited to a humanoid robot that uses artificial intelligence (AI). “A.I. God. Portrait of Alan Turing” (2024) was made by Ai-Da Robot, an artist robot that was the idea of Aidan Meller, an art dealer in Oxford, and might fetch up to $180K for what’s being described as “creepy ‘AI art’”. At least that’s what Aidam Meller hopes .
Sounds like great business for the art dealers and auctioneers. But what’s in it for the buyer? After all, it’s just AI-generated. Isn’t the whole idea to make it cheap and accessible? The New York Times questions whether it should even be allowed to profit from AI-generated art. Meanwhile, other artists are trying to adapt and work out how to survive and thrive using AI tools, with Adobe executives even warning that artists that don’t embrace AI will be left behind.
This inevitably begs the question: What do we actually pay for when we buy art? Is it the artistry, the workmanship, and the mastering of the trade? Or is it the artist's vision and imagination? Is it a fusion of both? And where does that leave us with AI, with it’s curious blend of prompt-driven symbolic ideation and digital image generation? "Game-changing," "revolutionary,” and “disruptive” are thrown around too carelessly, but in the case of AI image, video, and sound generation, the terms seem entirely appropriate.
Who’s Bias is it anyway?
We’ve spoken about the simple fact that it’s impossible to create “unbiased” AI before in Curious AI 33 and Curius AI 16 . This is simply because everyone has biases, and what people really mean when they say “unbiased” is that they want their own biases reflected.
Another study has just been released, titled “Large Language Models Reflect the Ideology of Their Creators," that examines how large language models (LLMs) reflect the ideologies of their creators. The researchers found significant normative differences in responses based on language and regional context.
If I were naive, I would say there are important implications here for regulatory frameworks and regulations aiming for ideological neutrality. In the meantime, back down here in the real world, logical analysis indicates that controlling AI model biases will become a major battleground in education, politics, business, and, as we’re already seeing in hiring.?
When reports talk about job hunters using AI tools to help get past the recruiters’ bots, what they are really saying is that the AI tools used by recruiters select candidates based on bias. Job seekers are using specialized tools that mirror these biases back to the recruiters models.?
Large Language Models Reflect the Ideology of Their Creators
Greenshoots or Grift and Hype?
This week’s reality check:
AI Transcription in Healthcare
Researchers are raising concerns about AI transcription tools used in healthcare. The tools reportedly produce inaccurate outputs due to hallucinations. Despite assurances from the developer that it’s fine-tuned for medical language, even OpenAI cautions against using such technology for critical transcriptions. Wired reports that hospitals are using it anyway???
Learn more about the latest Quantum Technology
Check out the Intriguing Quantum Newsletter .
AI Business
Perplexity Surges to 100 Million Weekly Search Queries Amid E-commerce Expansion
Perplexity has reached 100 million weekly search queries, increasing significantly in recent months. The company plans to introduce e-commerce features and an ad model, while facing legal challenges over content usage. A funding round is anticipated, aiming for a valuation of $8 billion.
techcrunch.com | https://techcrunch.com/2024/10/25/perplexity-says-its-now-serving-100m-search-queries-a-week/
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
Masayoshi Son Inflates The AI Bubble Even More
Masayoshi Son's extreme investments showcase his volatile approach to the AI sector, contributing to speculation in the market. His history of significant gains and losses indicates an unpredictable future for AI investments, potentially inflating the current AI bubble further.
bloomberg.com | https://www.bloomberg.com/opinion/articles/2024-10-25/masayoshi-son-inflates-the-ai-bubble-even-more
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI is a bubble
Elon Musk in funding talks with Middle East investors to value xAI at $45bn
Elon Musk is engaging with investors in the Middle East to secure funding aimed at positioning xAI's valuation at $45 billion. This move signals Musk's commitment to expand his AI ambitions while leveraging international investment opportunities.
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will evolve gradually
OpenAI CEO Sam Altman Warns Lack of Compute Is Delaying Product Launches
OpenAI's CEO, Sam Altman, acknowledged that insufficient computing capacity is hindering product releases. He highlighted technical complexities and limited resources as critical factors delaying the company's next AI models and features, emphasizing the importance of compute allocation amidst competing projects.
techcrunch.com | https://techcrunch.com/2024/10/31/openai-ceo-sam-altman-says-lack-of-compute-is-delaying-the-companys-products/
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
AI Carbon Footprint
GenAI's dirty secret: It's set to create a mountainous increase in e-waste
New research indicates generative AI (GenAI) could drastically increase e-waste by 2030, forecasting up to 2.5 million tons annually unless mitigation measures are implemented. This surge may result from the rapid deployment of AI servers, emphasizing the need for circular economy strategies to address rising electronic landfill issues.
Sentiment: Negative | Time to Impact: Mid-term | Future Signal: AI will accelerate
Just How Much Energy Does Generating An AI Image Actually Use?
A study by Hugging Face and Carnegie Mellon reveals that generating a single AI image can consume up to half a smartphone battery, equating to approximately 0.011 kilowatt hours. With daily AI image generation estimates at 34 million, the environmental impact has garnered political attention and prompted potential legislation for
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI Healthcare
Concerns about medical note-taking tool raised after researcher discovers it invents things no one said — Nabla is powered by OpenAI's Whisper
Researchers have raised concerns about Nabla, an AI transcription tool used in healthcare, which reportedly produces inaccurate outputs due to hallucinations. Despite assurances from the developer that it’s fine-tuned for medical language, OpenAI cautions against using such technology for critical transcriptions, emphasizing verification needs.
tomshardware.com | https://www.tomshardware.com/tech-industry/artificial-intelligence/concerns-about-medical-note-taking-tool-raised-after-researcher-discovers-it-invents-things-no-one-said-nabla-is-powered-by-openais-whisper
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI Trust, Risk, and Security Management
Japanese Man Convicted of Malware Creation Using Generative AI
A Tokyo court has convicted Ryuki Hayashi, 25, for developing malware with generative AI, sentencing him to a three-year suspended term. This case marks a precedent in judicial response to AI-driven cybercrime, highlighting the challenges of regulating AI misuse for unlawful purposes.
japantoday.com | https://japantoday.com/category/crime/japanese-man-convicted-of-creating-malware-using-generative-ai
Sentiment: Negative | Time to Impact: Short-term | Future Signal: AI will evolve gradually
Transcription tool powered by AI found to ‘hallucinate’ extra sentences
OpenAI's transcription tool, Whisper, exhibits a serious flaw by fabricating false sentences during transcription. Experts warn of potential risks in industries reliant on accurate text, particularly healthcare, prompting calls for stricter regulations and improvements to the tool to mitigate harmful inaccuracies.
bordercountiesadvertizer.co.uk | https://www.bordercountiesadvertizer.co.uk/news/national/24679860.transcription-tool-powered-ai-found-hallucinate-extra-sentences/
Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will evolve gradually
AI Warbots
ARX Robotics Seeks Series A Funding to Become Europe’s Next Defence Prime
ARX Robotics, a German startup specializing in unmanned ground vehicles, is pursuing Series A funding following a successful €9m seed round. With a mission to establish itself as Europe's leading defense provider, it emphasizes adaptability and modularity while addressing challenges in defense procurement and talent acquisition.
领英推荐
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
U.S. Military Secures First Confirmed OpenAI Purchase for Enhanced Combat Operations
The U.S. Africa Command has officially approved the first purchase of OpenAI's technology through Microsoft, aiming to enhance combat operations and data processing capabilities. This represents a significant step in integrating AI tools into military operations, marking a shift in defense procurement strategies.
Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will accelerate
AI at Work and Employment
LinkedIn launches its first AI agent to take on the role of job recruiters
LinkedIn has unveiled its first AI agent designed to enhance the recruitment process by emulating job recruiters. This innovative tool aims to streamline hiring and improve candidate matching, positioning LinkedIn at the forefront of AI-driven recruitment solutions.
techcrunch.com | https://techcrunch.com/2024/10/29/linkedin-launches-its-first-ai-agent-to-take-on-the-role-of-job-recruiters/
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
AI in Art and the Media
Forget AI image generators, an autonomous AI artist just made $351,600 at Sotheby's
Botto, an autonomous AI artist, generated $351,600 in sales at Sotheby’s, underpinning a significant trend in AI art. Backed by a community for NFT selection, Botto raises questions about traditional authorship, indicating a potential shift in the art market's landscape.
creativebloq.com | https://www.creativebloq.com/ai/ai-art/forget-ai-image-generators-an-autonomous-ai-artist-just-sold-usd351-600-in-paintings-at-sothebys
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI is hype
Adobe execs say artists need to embrace AI or get left behind
Adobe executives assert that artists must adopt generative AI tools to ensure future success, warning that reluctance to embrace such technology may result in obsolescence. The company aims to meet growing demand for efficient creative processes while navigating strong criticism from traditional artists.
theverge.com | https://www.theverge.com/2024/10/25/24278715/adobe-artists-embrace-generative-ai-creative-community
Sentiment: Neutral | Time to Impact: Short-term | Future Signal: AI will accelerate
AI in Journalism
Meta Partners With Reuters to Enhance AI Chatbot's News Capabilities
Meta collaborates with Reuters to integrate news content into its AI chatbot, enabling the provision of news information and citations. This multi-year agreement signals a shift in Meta's approach towards news, amid ongoing debates on compensating news sources for content used by tech companies.
theverge.com | https://www.theverge.com/2024/10/25/24279259/meta-reuters-ai-chatbot-deal-news-licensing-media
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will evolve gradually
AI Slop is Flooding Medium
Medium grapples with a surge in AI-generated posts, with estimates suggesting over 40% of recent content may be machine-made. Despite a profitable summer, leadership acknowledges rising concerns about content quality and reader engagement, as they strive to balance human curation against the deluge of automated submissions.
Sentiment: Negative | Time to Impact: Immediate | Future Signal: AI will accelerate
AI in Science
Simple science summaries written by AI help people understand research and trust scientists
AI-generated summaries significantly enhance public comprehension of scientific research, outpacing human-written counterparts. These simplified texts bolster trust in scientists, highlighting AI's potential to transform science communication and restore public engagement in scientific discourse. Effective simplification is vital in advancing scientific literacy and trust.
theconversation.com | https://theconversation.com/simple-science-summaries-written-by-ai-help-people-understand-research-and-trust-scientists-241105
Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate
AI in Software Development
Over 25% of Google’s code is now written by AI—and CEO Sundar Pichai says it’s just the start
Google reports that over 25% of its new code is now generated by AI, indicating a significant shift in software development. CEO Sundar Pichai stresses that this is only the beginning, showcasing the potential of AI in enhancing productivity and supporting business growth.
Sentiment: Positive | Time to Impact: Immediate | Future Signal: AI will accelerate
AI in the Data Center
Energy exec punts datacenter power options out to long term
Schneider Electric's Steven Carlini emphasizes the importance of diversifying energy sources for datacenters, advocating short-term gas turbine solutions and long-term investments in small modular reactors. The quest for reliable power reflects the increasing demand amid technological growth, particularly in AI and cloud services.
Sentiment: Neutral | Time to Impact: Mid-term | Future Signal: AI will accelerate
Datacenter Developer Says Power Issues Holding Up New Builds
UK datacenter developer Segro faces significant delays in new builds due to power access issues, highlighting a global trend impacting datacenter capacity. This situation hinders investment in critical infrastructure, as the demand for energy from datacenters grows, particularly driven by AI and cloud services.
Sentiment: Negative | Time to Impact: Mid-term | Future Signal: AI will evolve gradually
Overcoming operational inertia in the race for AI-driven data centers
As global interest in AI surges, data centers are vital for economic growth. With projected spending of 24% increase in 2024 and $22 billion invested in early 2024, operators face challenges from operational inertia, power access, and ESG concerns, inhibiting their ability to capitalize on this demand.
techradar.com | https://www.techradar.com/pro/overcoming-operational-inertia-in-the-race-for-ai-driven-data-centers
Sentiment: Positive | Time to Impact: Short-term | Future Signal: AI will accelerate
The Path to AGI
Reclaiming AI as a Theoretical Tool for Cognitive Science
The article argues for a shift in the perception of AI from a practical tool for replicating human cognition to a theoretical framework useful in cognitive science. The authors emphasize the computational intractability of modeling human cognition, cautioning against the current AI hype.
Sentiment: Negative | Time to Impact: Long-term | Future Signal: AI is hype
Softbank CEO Predicts 'Super AI' Arrival by 2035, Projected Cost of $9 Trillion
Softbank's CEO Masayoshi Son forecasts the emergence of a "super AI" by 2035, potentially boosting global GDP by 5%. He estimates a need for $9 trillion in investments and unprecedented power requirements, positioning Nvidia as key in this future landscape.
Sentiment: Hyperbolic | Time to Impact: Long-term | Future Signal: AI is hype
Want to discuss AI, quantum, and other emerging technologies?
Join our Slack
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.