Mediaeater Digest Vol 30, No. 144

Meta AI chief says large language models will not reach human intelligence. (FT) At an event in London on Tuesday, DeepMind’s chief Sir Demis Hassabis said what was missing from language models was “they didn’t understand the spatial context you’re in . . . so that limits their usefulness in the end”.

Cars & Consumer Data: On Unlawful Collection & Use  (FTC)  Car manufacturers—and all businesses—should take note that the FTC will take action to protect consumers against the illegal collection, use, and disclosure of their personal data. Recent enforcement actions illustrate this point:

Visibility into AI Agents (arxiv) Visibility informs actions to manage risks from the deployment of increasingly agentic systems, but is not by itself sufficient. Even with a comprehensive understanding of agent activities, those harmed by them may not have the power to intervene and reduce risks. To best make use of visibility, future work could investigate increasing public influence over AI development and deployment,  developing a wide range of potential policy levers, and implementing infrastructure and practices to prevent or defend against harms

News Corp and OpenAI today announced a multi-year agreement. So far, Axel Springer has secured a $30 million, three-year deal, while the Financial Times landed a $10 million per year agreement. Rupert Murdoch has netted a substantial rate of $50 million per year, but this covers a bundle of at least seven publications. OpenAI is currently offering between $7 million to $10 million per year per publication as the prevailing rate in the industry. For more context, it would take an 100’s of thousands of deals to satisfy what was stolen. Lastly, getting in early is crucial; the last one in line will face the least favorable deal terms. Lastly if you were an IP rights holder suing them…. Also no word on how Rupert might integrate this with his spy craft. 

DOJ Ticketmaster  (DOJ). Live and direct at 11 AM  EDT    Also DOJ more music news,  Purported Rap Promoter Pleads Guilty to Role in $2.3M Nationwide Fraud Conspiracy.  (ouch “purported”). 

DOJ and Ticketmaster: A Timeline
– 1976: Ticketmaster is founded.
– 1991: Ticketmaster merges with Front Line Management; DOJ investigates and allows the merger with conditions.
– 2010: Ticketmaster merges with Live Nation to form Live Nation Entertainment; DOJ approves with conditions to foster competition.
– 2019: DOJ launches an investigation into Live Nation Entertainment for potential violations of the 2010 merger conditions.
– 2023: January: Senate Judiciary Committee holds a hearing on competition in the ticketing industry, fueled by controversy surrounding Ticketmaster’s handling of Taylor Swift’s tour tickets.
–  May 23 -2024  DOJ files a new lawsuit against Live Nation Entertainment, alleging antitrust violations and anti-competitive practices “It is time to break it up”.

On device

When Google announces it can monitor phone conversations for hostile scams, and Microsoft claims it can search everything you’ve ever touched on your computer, the importance of on-device AI becomes clear. Apple is likely to unveil their own on-device AI capabilities at their upcoming developer event, solidifying this trend.

The question is, how much of this AI processing will truly stay on-device? Will learnings, user preferences always remain local? It’s crucial to understand the implications of on-device AI for personal OPESEC.

On-device AI can be more nuanced.  Some examples,,,

  1. Model Updates: While the AI model itself runs on your device, updates to that model might be downloaded from external servers. These updates could include new features, improved accuracy, or bug fixes.

  2. Anonymized Data: Some on-device AI systems may collect anonymized data about usage patterns or performance to help developers improve the model. This data is stripped of personally identifiable information before being sent.

  3. User Consent: Certain applications might request your permission to share specific data for specific purposes, even with an on-device AI model. For example, a voice assistant might ask to send your voice commands to a server for better speech recognition.

  4. Hybrid Models: Some AI systems use a combination of on-device and cloud-based processing. In these cases, certain parts of the AI model might run on your device, while others rely on external servers for more computationally intensive tasks.

Whats possible now(or what should always be local, and if not, ask why?)

Optimist in me says –  Most important trend done correctly, solves for latency, keeps works offline, keeps data on-device

Pessimist in me says:  Privacy is always the first victim of  convenience. Creep happens, and this feature creep will lead to more and more user consent in the guise of utility and convenience likely geeting neither in the short term.

I came across this post from Clem at Hugging Face, it summed it up nicely while pointing to four models to use . “- No cloud, no cost, no data sent to anyone, no problem. Welcome to local AI on Hugging Face! ”  Open source FTW

Mike Kelly
Mike Kelly

MED Vol.30, No. 139

Songs I listened to compiling this:
Quarters of Change – Turn It Away  (youtube)
Rosie Tucker – All My Exes Live In Vortexes (youtube)
Remi Wolf – Cinderella (youtube)
Bottle Rockets – Community Service (youtube)

The race for an AI-powered personal assistant  (FT)  “In the new future every single interaction with the digital world will be through an AI assistant of some kind. We will be talking to these AI assistants all the time. Our entire digital diet will be mediated by AI systems,” (editors note: assistant’s are gateways to twinning. You don’t own your digital twin)

Culture secretary Lucy Frazer says the UK is working on rules around the use of creative work for training AI after media and arts executives voiced concerns (FT) Frazer said ministers would initially focus on ensuring greater transparency over what content was being used by AI developers to train their models, which will in effect allow the industry to see if work it produces is being ripped off.

To the journalists contacting me about the AGI consensual non-consensual (cnc) sex parties— (@soniajoseph) There is nothing wrong imo with sex parties and heavy LSD use in theory, but combined with the shadow of 100B+ interest groups, leads to some of the most coercive and fucked up social dynamics that I have ever seen. The climate was like a fratty LSD version of 2008 Wall Street bankers, which bodes ill for AI safety.

Hollywood at a Crossroads: “Everyone Is Using AI, But They Are Scared to Admit It” (HR) Some sectors of the industry are already threatened with extinction. “Dubbing and subtitling employment in Europe is finished,” says Kavanagh. ./. Over the next three years, it is estimated that nearly 204,000 positions will be adversely affected. Concept artists, sound engineers, and voice actors stand at the forefront of that displacement. Visual effects and other postproduction work were also cited as particularly vulnerable.

Declaration of AI Training Opt Out  (sonymusic) SMG’s affiliates, Sony Music Publishing (SMP) and Sony Music Entertainment (SME), on behalf of themselves and their wholly owned or controlled affiliates, are making this affirmative, public declaration confirming that, except as specifically and explicitly authorized by either SME or SMP, as the case may be, each of them expressly prohibits and opts out of any text or data mining, web scraping or similar reproductions, extractions or uses (“TDM”) of any SME and/or SMP content (including, without limitation, musical compositions, lyrics, audio recordings, audiovisual recordings, artwork, images, data, etc. 

New A.I. Works Fuse Fantasy and Memory at Photo London For those hoping to determine what new directions the medium might be taking, the fair is littered with pieces that show how photography keeps evolving as it feeds into and emerges out of all manner of artistic practices. On show are textiles, sculptural pieces, and paintings, alongside an increasingly prominent yet still controversial presence: A.I. Here are some of the most interesting examples on show.

Director Quentin Dupieu’s ‘The Second Act’ debuted at Cannes a a movie-within-a-movie about making movies.  Léa Seydoux, Vincent Lindon, Louis Garrel and Raphaël Quenard play actors making the first film directed by AI.

The Collapse of the News Industry Is Taking Its Soul Down With It (Politico)The decline of the media is sapping journalism of a crucial tool. “For the next generation, the future is dismal. And the youth know it. With an eye on the shrinking industry, fewer and fewer students now collect bachelor’s degrees in journalism, according to numbers collected by MIT Media Lab’s DataUSA site. Sensing the trend, schools like the University of New Hampshire, Ohio Wesleyan, Arkansas State University and others have ended their journalism programs in the last academic year or cut way back. “

 In this weeks ‘the worst people’ section 

I’m the new Oppenheimer!’: my soul-destroying day at Palantir’s first-ever AI warfare conference (The Guardian)  But Karp, who’s known as a provocateur, aggressively condoned violence, often peering into the audience with hungry eyes, palpably desperate for claps, boos or shock. He began by saying that the US has to “scare our adversaries to death.   

 

 

read or reading

MED Vol.30, No.132

Random notes…

Music is always first, always. 

Surviving the first wave of digital disruption on the first beachfront gave me the tools needed to understand shifts across all sectors, with web1, web2 and now three.  The learnings in the arbitrage of disruption of music provided super digital powers for those who did not fight, but embraced and reshaped, though the insights from hand to hand combat.

Our music model @elevenlabsio  is coming together! Here’s a very early preview. Here we have full lyrical generation at full song length. The accelerators dropping each day in this category that move the creative process further downstream is astonishing,  It’s no wonder Lucian Grange TikTok deal points for UMG had AI in the mix, I look forward to seeing the enforcement mechanisms. 

Current state of the internet:  Large and small companies working on solving for understanding individual indent by inferring it though process like tutoring and learning.

We will see prompting fade as intent ascends, some LLMs are already doing this like Anthropic. By empirical observation the tacit understanding that search was search-and- destroy your thought process through multiple options. We lived like that for two decades. The alliteration of words that start with C have now added words like ‘context’ to the mix.

 My head hurts.  Some recent links

Economic desperation drives 500,000 Argentinians to Sam Altman’s Worldcoin crypto project (techspot) Worldcoin, the iris-scanning crypto project from OpenAI CEO Sam Altman, is finding plenty of success in Argentina, where 500,000 people have handed their biometric data over to the company. But with inflation in the Latin American nation at around 288% and Worldcoin offering $50 in crypto to everyone who looks into its Orbs, the figure comes as little surprise.

WPP boss targeted by deepfake scammers using voice clone  (ft) Mark Read, chief executive of WPP, has been the target of a deepfake scam in which criminals used a voice clone and public YouTube footage to set up a video meeting with his executives. In an email to colleagues seen by the Financial Times, Read said that the scammers used a publicly available photo to set up a fake WhatsApp account under his name.

Congressional Committee Threatens To Investigate Any Company Helping TikTok Defend Its Rights (techdirt) Members of Congress threatened to investigate NetChoice if it didn’t drop TikTok from its roster. Politico had some of this story last night, claiming that there was pressure from Congress to drop TikTok:

OpenAI Model Spec  (openai) We believe developers and users should have the flexibility to use our services as they see fit, so long as they comply with our usage policies. We’re exploring whether we can responsibly provide the ability to generate NSFW content in age-appropriate contexts through the API and ChatGPT. We look forward to better understanding user and societal expectations of model behavior in this area.   We’re streaming live at 10AM PT on Monday, May 13 to demo some ChatGPT and GPT-4 updates.

Apple Nears Deal With OpenAI to Put ChatGPT on iPhone (bloomberg) Move is part of push to add AI features to Apple devices Company also held talks with Google about using Gemini

As states take on AI, tech’s top lobbyist fights back Q: How much are you and people in Washington focused on the election and possible AI involvement in election interference? A: People are worried about it. But it won’t be until something pretty significant happens and that gets a ton of attention, that the general public will focus on it a lot. Of course, we’re trying to make sure that doesn’t happen.

Randy Travis sings again, courtesy of AI  (cbs) In 2013 country singer Randy Travis suffered a massive stroke, which paralyzed his right side and damaged an area of his brain that controls speech and language. He has not quite recovered the ability to sing, but working with singer James DuPre and a computer program that creates an AI-generated version of his voice, Travis and his longtime producer Kyle Lehning have created a new song, “Where That Came From,” that captures Randy’s country heart. Lee Cowan reports.    — at last happy use cases…

AI Doesn’t Threaten Humanity. Its Owners Do  (scientific america) In fact, that information harvesting is the whole business model of our digital and smart-device-enabled world. All of our habits and behaviors are monitored, reduced to “data” for machine learning AI, and the findings are used to manipulate us for other people’s gains. It doesn’t have to be this way. AI could be used more ethically for everyone’s benefit. We shouldn’t fear AI as a technology. We should instead worry about who owns AI and how its owners wield AI to invade privacy and erode democracy.

Introducing recon – a CLI tool to gather context for LLMs But one problem I’m constantly running into is that there are things I’d like to ask of an AI, but either there is a ton of data or it’s spread all over the place. For example, if I’m working on RPG Portrait, I may want to discuss growth rates and have it help me prioritize my tasks accordingly. But the growth rates are in a Postgres database, the tasks are in GitHub issues, and it probably needs a bit of the app’s source code too. That’s where recon comes in.

Oscars launch fundraising drive as Academy Awards audience shrinks (ft) The Academy of Motion Picture Arts & Sciences is launching a $500mn global fundraising drive to bolster its finances and counter a decades-long decline in the audience for the annual Oscars TV broadcast that generates most of its revenue.  “The Academy is acknowledging the fact that we’re living in a much more globally focused world, a much more globally connected world,”

(Legacy media realizing they have perished – ED)

The Unwanted Shelter Dog Who Found His Way to Westminster. (nyt)

 

MED = Mediaeater Digest publishing links since 1993

DHS / AI and open-source technology

Over 20 Technology and Critical Infrastructure Executives, Civil Rights Leaders, Academics, and Policymakers Join New DHS Artificial Intelligence Safety and Security Board to Advance AI’s Responsible Development and Deployment (DHS April26, 2024)

The executives break down like this…

Software and hardware companies: Sam Altman, CEO, OpenAI; Dario Amodei, CEO and Co-Founder, Anthropic; Jensen Huang, President and CEO, NVIDIA; Arvind Krishna, Chairman and CEO, IBM; Fei-Fei Li, Ph.D., Co-Director, Stanford Human-centered Artificial Intelligence Institute; Satya Nadella, Chairman and CEO, Microsoft; Shantanu Narayen, Chair and CEO, Adobe; Sundar Pichai, CEO, Alphabet; Adam Selipsky, CEO, Amazon Web Services; Dr. Lisa Su, Chair and CEO, Advanced Micro Devices (AMD)

Critical infrastructure operators: Ed Bastian, CEO, Delta Air Lines; Vicki Hollub, President and CEO, Occidental Petroleum

Public officials: Bruce Harrell, Mayor of Seattle, Washington; Chair, Technology and Innovation Committee, United States Conference of Mayors; Wes Moore, Governor of Maryland

Civil rights community: Rumman Chowdhury, Ph.D., CEO, Humane Intelligence; Alexandra Reeve Givens, President and CEO, Center for Democracy and Technology; Maya Wiley, President and CEO, The Leadership Conference on Civil and Human Rights

Academia: Fei-Fei Li, Ph.D., Co-Director, Stanford Human-centered Artificial Intelligence Institute; Nicol Turner Lee, Ph.D., Senior Fellow and Director of the Center for Technology Innovation, Brookings Institution

What’s missing is OPEN-SOURCE representatives.  They seemed to be aligned with using open-source technology , In March (24) DHS announced they would be using open source software as part of their pilot programs announced:

Transform Security Investigative Processes, Unlock Data-Driven Insights, and Improve Mission Outcomes – HSI’s pilot project will strengthen their investigative processes by introducing a LLM-based system designed to enhance the efficiency and accuracy of summaries investigators rely upon. The LLM-based system will leverage open-source technologies to allow investigators to more quickly summarize and search for contextually relevant information within investigative reports. The pilot could lead to increases in detection of fentanyl-related networks, aid in identification of perpetrators and victims of child exploitation crimes, and surface key patterns and trends that could further HSI’s vital work.

Related:

DHS AI Roadmap

opensourceinitiative/osaid-0-0-8

SB-1047 will stifle open-source AI and decrease safety

LLM’s From Skill to Proficiency 

Early on, LLMs had a reputation for laziness, often stopping short of a complete answer or list. But that final 10% is crucial – it’s the difference between ‘just okay’ and truly effective, between a rough idea and a tool you can actually use

Skill: A developed ability to do something well.
Knowledge: Theoretical or practical understanding of a subject.
Aptitude: A natural ability or talent for acquiring skills and knowledge within a domain.
Mastery: Comprehensive or in-depth skill and knowledge in a particular area, indicating a high level of proficiency.
Proficiency: The level of competence in a skill or area of knowledge.

We have good performance in complex language understanding tasks, language translation, customer support automation and dare I say LLM’s excel at creative content generation.  That said, I am still finding it hard to say we have continuous proficiency 

Highlighting what’s missing in an “almost” answer, LLM’s call it a feature and it can help us learn or prompt us to think about something in new ways, this applies well to creative applications but not so much for the other proficiencies listed above. 

The unspoken truth right now is that LLMs cannot provide 100% proficiency at any task consistently.  It is very hard to guarantee complete accuracy or reliability.  LLM’s are probabilistic not authoritative.  Factual accuracy, nuanced context and originality in reasoning are  all less then ideal as they are remixes and representing existing information. 

I believe that LLM’s will remain tools to annotate and augment human domain expertise but never replace it.  There will be improvements through fine tuning and continuous learning and the eventual combining of reasoning tools to better achieve factual accuracy.