FT does a video a/b test w/SORA and Runway and Pika labs (FT)
MED Vol.30, No.121
News
Apple must open iPadOS to sideloading within 6 months, EU says (arstech)
Will An AI Movie Win An Oscar Within Seven Years? (forbes)
Industry
Google Quantam AI (google)
Google AI Essentials. (google edu course)
Application/Tools
Gpt2-chatbot @ LMSY (details)
Data Bonsai: a Python package to clean your data with LLMs (github.com)
NIST Press release / GenAI – Evaluating Generative AI. (nist)
Anon.com. (ai agents)
Cool
OH at @USV
The Simpsons – 1950’s Super Panavision 70
Horror Show
Work assignment algorithms at Uber and others appear to be far more more intrusive of worker privacy than I had ever imagined
FCC fines AT&T, Sprint, T-Mobile, and Verizon for illegally sharing location data
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:
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.
MED Vol.30, No.117
Survey reveals translators and illustrators losing work to AI (societyofauthors)
FCC Restores Net Neutrality (fcc)
Start Your Own ISP (startyourownisp)
A look at the early impact of Meta Llama 3 (meta)
AI Factory for the New Industrial Revolution – Nvidia GTC24 (youtube)
One Bad Apple Can Spoil Your IPv6 Privacy (arxiv)
Overclocking My Brain: Adapting to Faster Audio and Reading Speeds (buttondown)
Apple releases eight small AI language models aimed at on-device use (arstechnica)
Watch it and weep (or smile): Synthesia’s AI video avatars now feature emotions (techcrunch)
Alphabet 2024 Q1 Earnings Call (abc.xyz) (nice domain)
OpenVoice: Instant Voice Cloning (github)
MED Vol. 30, No. 116
Personal Explorations in Distributed Cognition with Foundation Models (medium)
Meta Platforms, Inc. (META) First Quarter 2024 Results Conference Call (meta/pdf)
Augment (Augment) Eric Schmitt’s GIT alternative
VideoGigaGAN: Towards detail-rich video super-resolution (github)
An AI startup made a hyperrealistic deepfake of me that’s so good it’s scary (mit tech review)
When Do We Stop Finding New Music? A Statistical Analysis (statssignificant) TL/DR 30 years old
Related: new music from this year: Lucy Rose – Life’s Too Short (Lyric Video) , Maggie Rogers – Don’t Forget Me (Official Video)
FM Skyline : Images ; Blue Lab Beats
How TikTok’s Chinese owner tightened its grip on the app (ft)
The Meta-morphosis of Mark Zuckerberg (nyt) Venessa Friedman breaks down the vibe shift
Reid Hoffman meets his AI twin – Full (youtube)
The effect of speed limits on accident frequency on the German Autobahn (pdf) TL/DR slowing to 75MPH saves lives
Rule Requiring Automatic Refunds of Airline Tickets and Ancillary Service Fees (transportation gov)
The Six Horsemen of the Digital Apocalyps (iccl.ie)
De-DSI is a proof-of-principle of fully decentralised search engines. (hf) Arxiv paper (pdf) TL/DR decentralized AI-powered search
MED Vol. 30, No. 110
We announce the successful editing of DNA in human cells with gene editors (x post)
Generative A.I. Arrives in the Gene Editing World of CRISPR (nyt)
Introducing Our Open Mixed Reality Ecosystem (meta) tactic, to this strategy
The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions (huggingface)
How the BBC used face swapping to anonymize interviewees Case study PDF (BBC)
Spotify Q1 2024 Shareholder deck (pdf)
FineWeb: 15 trillion tokens of high quality web data. (huggingface)
Meet Your New Assistant: Meta AI, Built With Llama 3 Meta AI experience (meta)
LLM Scraper (github)
Hollywood’s AI Concerns Present New and Complex Challenges for Legal (variety)
MED Vol.30, No.109
The Life and Death of Hollywood
High-speed trader Jane Street raked in $4.4bn at start of 2024
Americans Deserve More Than the Current American Privacy Rights Act
Block the “Everyone is a spy” surveillance bill
NSA Publishes Guidance for Strengthening AI System Security
OpenAI. They have a spider problem
Innovative Nuca-Cam Challenges Privacy Norms in the Age of AI
Inference + Intent
AIML – Inference and intent
intent. – 1 : a usually clearly formulated or planned intention : AIM the director’s intent
2 a : the act or fact of intending : PURPOSE ; b : the state of mind with which an act is done : VOLITION
Inference – A conclusion reached on the basis of evidence and reasoning. “researchers are entrusted with drawing inferences from the data”
Intent and inference have long been integral to business processes. AI/ML provides new opportunities as illustrated in my hand drawn diagram above. The application of AI/ML to infer intent, which is then used to personalize customer experiences and enhance operational efficiency.
Google is realizing LLM’s infer intent easier, simply by the nature of the query. This poses a systemic challenge for Google to address stat (Gemini), as it affects the business’s flywheel dynamics and illustrates intent in action.
Trend: Pairing this with Theory of mind. Theory of mind (ToM) is the (alleged) ability to understand and reason about other people’s mental states. This includes understanding their beliefs, desires, intentions, and emotions. It also involves using this information to explain and predict human behavior.
What could go wrong? Emergent (actual) example: agents that have agency, inferring via ToM your intent and taking autonomous action. There is the fact that intent is not action, and inferring is not understanding. Put another way, To plan is not to do. inference invites errors.