Mr. ChatGPT and Generative AI Go to Washington

Welcome to the 9th issue of the AstroFeather AI newsletter!

It's been another big week for AI! The news headlines were dominated by discussions about the recent US Senate hearing for oversight and regulation of AI. There were also some exciting research results for the use of AI in the early detection of Parkinson’s Disease, using NeRFs to create photorealistic avatars, and some trending product launches and announcements.

I hope you enjoy reading this week’s updates and if you have any helpful feedback, feel free to respond to this email, contact me directly [ LinkedIn ], and follow the AstroFeather LinkedIn!

Thanks - Adides Williams, Founder @ AstroFeather

[ Daily AI News | Newsletter Signup | Past Newsletters | AstroFeather LinkedIn ]

In today’s recap (10 min read time):

• Mr. ChatGPT Goes to Washington.

• (Research) AI Platforms for Early Detection of Parkinson’s Disease and Creating Photorealistic Full-Body 3D Avatars.

• Product Previews and Launches.

• Company Announcements and News Throughout the Industry.

Must-Read News Articles and Updates

Update #1. Mr. ChatGPT Goes to Washington.

The latest: Sam Altman, CEO of OpenAI (known for creating ChatGPT and GPT-4)), recently testified during a three-hour Senate hearing where he called for regulation to guide the development of AI, proposed the creation of a new government agency to regulate AI companies (through model licensing and testing requirements), and addressed concerns about the potential risks associated with the development of advanced AI systems.

How it works: In general, a U.S. Senate hearing is a meeting or session during which committee members conduct investigations, evaluate the activities of a government agency, or obtain information and expert testimony to assist in the creation of new laws (legislation).

On Tuesday, May 16th, 2023, the hearing of the Senate Judiciary Subcommittee on Privacy, Technology, and the Law was presided over by Chairman and U.S. Senator Richard Blumenthal (D-CT). Several members of Congress were present to gather information from expert witnesses during what Senator Blumenthal described as “first in a series of hearings intended to write the rules of AI.”

Experts who appeared before the Senate Judiciary Committee to provide testimony and opinions included Sam Altman (CEO of OpenAI), Christina Montgomery (Chief Privacy & Trust Officer at IBM), and Gary Marcus (Professor Emeritus of Psychology and Neural Science at New York University).

What they discussed: During the Senate committee meeting, all of the witnesses received a warm welcome from the lawmakers, perhaps because the three agreed that AI should be regulated, and quickly. Here are some key takeaways from the discussion:

[No. 1] All witnesses support AI regulation:

• Sam Altman (OpenAI): During the hearing, Altman laid out a three-point plan that includes creating a new government agency that can license (or revoke) AI models "above a certain threshold of capabilities," creating safety tests and standards for AI models that must be passed before the model can be released to the public, and requiring independent audits (by independent experts) of an AI model's performance.

• Christina Montgomery (IBM): Montgomery argued for "precision regulation." She mentioned the need to "define the highest risk uses of AI," create rules specific to different use cases of AI, and lean on existing government agencies to act now rather than waiting for a new AI-focused regulator to be created.

• Gary Marcus, professor emeritus, New York University: Marcus advocated for "close collaboration between independent scientists and governments" to ensure that scientists can help evaluate advanced AI systems before they are released to the public.

[No. 2] Policymakers from both parties share concerns about the potential impact of AI. Some of the concerns raised include:

• Election misinformation and its impact on voters

• Impersonation of public and private figures

• Disruption of jobs and economic displacement

• Copyright and licensing issues

• Distribution of dangerous and harmful content

[No. 3] Global regulators (EU and China) are leading the US in AI regulation:

• The European Union: After two years of negotiations, EU lawmakers are moving quickly toward a final draft of their AI Act, making it one of the first comprehensive pieces of AI legislation in the world. The EU AI Act includes rules for chatbots (such as ChatGPT) and aims to classify AI systems according to four levels of risk, with a focus on guarding against AI threats to health and safety while protecting fundamental rights, values, and vulnerable groups.

• China: The Cyberspace Administration of China's (CAC) new laws against "deepfake" technology took effect in January this year, requiring AI-generated content to be watermarked and comply with security regulations. Chinese officials also recently completed a second round of drafting regulations for the research, development, and use of generative AI (GenAI) applications and platforms.

[No. 4] Concentration of power and regulatory capture:

• Sen. Cory Booker (D-NJ) and other committee members expressed concern about what Booker called the "massive corporate concentration" of AI power in the hands of a few companies, specifically citing the multibillion-dollar OpenAI/Microsoft alliance.

• Gary Marcus also told lawmakers during the Senate hearing that a separate AI regulatory agency could fall victim to regulatory capture, a term used to describe when government agencies begin to serve the interests of the industries (or companies) they're supposed to regulate.

[No. 5] Lack of balance:

• Prominent AI ethicist Timnit Gebru expanded on the "concentration of power" concern, saying, "if industry writes the laws, then we have no balance." Indeed, this echoes Gary Marcus's comments at the Senate hearing, when he called for “tight collaboration between independent scientists and governments,” as well as Senator Blumenthal's opening remarks that “you can create 10 new agencies, but if you don’t give them the resources, and I’m talking not just about dollars, I’m talking about scientific expertise, [industry] will run circles around them.”

Additional Links for “Mr. ChatGPT Goes to Washington”:

• Regulations News Headlines

• Generative AI News Headlines

• ChatGPT News Headlines

Update #2 (Research): AI Platforms for Early Detection of Parkinson’s Disease and Creating Photorealistic Full-Body 3D Avatars.

Early detection of Parkinson's disease with CRANK-MS: Parkinson's disease is the second most common neurological disorder (after Alzheimer's disease), is spreading faster than any other neurological disorder, and is characterized by a progressive decline in movement, starting with difficulty walking and shaking of the limbs and face (in its early stages) and often leading to cognitive decline (in its later stages).

While early detection of Parkinson's disease is a challenge, a group of researchers at the University of New South Wales has developed an AI-based system called CRANK-MS that they claim can detect the underlying signs of Parkinson's years before symptoms appear. The machine-learning model uses blood plasma samples to identify patterns of certain chemical compounds, called metabolites, that can predict the presence of the disease or protect against it.

Study: The research team used blood plasma samples collected as part of the Spanish European Prospective Study on Nutrition and Cancer to identify patterns in metabolites. They focused on 39 patients who developed Parkinson's disease within 15 years of participating in the study, and compared their metabolite mixtures with those of 39 control patients who did not develop the disease.

Observations: The team identified several metabolite patterns that they considered potentially significant, including lower levels of triterpenoids in people who later developed Parkinson's. Triterpenoids deal with stress at the cellular level and are found in foods such as apples, olives, and tomatoes.

Moreover, the research team found the presence of polyfluorinated alkyl substances (PFAS) in people who went on to develop Parkinson's, which may also be linked to higher exposure to industrial chemicals. However, larger numbers of patients are needed for more conclusive studies.

Results: According to the research group, CRANK-MS was able to detect the risk of Parkinson's disease with up to 96 percent accuracy and identify chemical markers critical for predicting the development of the disease.

[ Research Paper | GitHub Code ]

Photorealistic Full-Body 3D Avatars with HumanRF: Avatars (digital representations of a user or a user's persona) have been widely used for decades, appearing most commonly in video games, online forums, non-gaming online worlds (such as Second Life), and social media platforms. As a result, avatar generation has proven to be a lucrative endeavor, with apps such as Lensa AI skyrocketing in popularity after the release of its "Magic Avatars" feature, which reportedly generated $29 million in in-app purchases 12 days after the app's launch (in December 2022).

While most avatar generators have focused on 2D facial images (such as profile pictures) or 3D cartoon-like characters, generating photorealistic full-body 3D avatars remains a challenge. Recently, a group of researchers at AI startup Synthesia successfully developed a method for generating high-resolution 3D avatars of people in motion called HumanRF.

HumanRF uses a machine learning technique that I shared with you in AstroFeather Newsletter Issue No. 5 called Neural Radiance Fields (NeRFs), a technique that has recently gained popularity for its ability to stitch together a series of 2D images into 3D scenes.

Perhaps most interestingly, the research team trained HumanRF on their own dataset, ActorsHQ, which consists of tens of thousands of frames of dynamic human motion captured at high resolution (12 megapixels) using multi-view video input, allowing playback from novel, unseen viewpoints.

[ Research Paper | Project Page | Video ]

Additional Links for “AI Platforms for Early Detection of Parkinson’s Disease and Creating Photorealistic Full-Body 3D Avatars”:

• AI Research and Development News Feed.

• AI Healthcare News Headlines.

Update #3. Product Previews and Launches.

ChatGPT is Officially on Mobile Devices: OpenAI has launched an official iOS app for ChatGPT. The free app allows users to interact with the chatbot and ask questions without a web search. It also features voice input and syncs with ChatGPT's web interface history.

ChatGPT Plus subscribers can access GPT-4 features through the app and receive early access to features and faster response times. The app is initially available for US users only, but will be rolled out to other countries in the coming weeks, with an Android version also in the works.

Sanctuary AI Humanoid Robot: Vancouver-based Sanctuary AI has unveiled its Phoenix humanoid robot. The 5'7” bipedal robot weighs 155 pounds, can lift objects up to 55 pounds, and travels at three miles per hour. Phoenix has complex hands with 20 degrees of freedom, giving it human-like dexterity.

In March, Phoenix was deployed in a retail store where it successfully completed more than 100 retail-related tasks. To date, Sanctuary AI has raised more than $100 million in funding from various sources, including a Series A round and the Canadian government.

Apple Improves Accessibility Features with AI: Apple is introducing a new set of accessibility features to its devices that use AI. The company's latest announcement includes several features such as an AI detection feature that can help people with vision issues, and an AI-based Point and Speak feature in the Magnifier app that reads text users point to.

Apple is also introducing features for users at risk of losing their ability to speak, including the Personal Voice app, which creates voice clones for users, and Live Speech, which converts text to speech.

Additional Links for “Product Previews and Launches”:

• Product Launch News Feed.

• Robotics News Headlines.

• Apple News Headlines.

• ChatGPT News Headlines.

Update #4. Company Announcements and News Throughout the Industry.

Rooms.xyz and 3D Spaces: is a new digital creativity platform developed by former Google employees (and backed by a16z) that provides users with a browser-based tool to create 3D spaces called “rooms” using drag-and-drop editable objects or code.

Inspired by a combination of factors, including the difficulty of creating 3D models, the platform aims to provide a middle ground between simple creation tools like Minecraft and more advanced world-building platforms like Roblox. Rooms comes with 1,000 custom voxel 3D objects that can be added, edited, and remixed within the platform.

Hippocratic AI for Healthcare: Hippocratic AI has raised $50 million in seed funding to develop a safety-focused large language model (LLM) designed specifically for healthcare. The technology is a text-generating model aimed at applications such as explaining benefits and billing, providing dietary advice and medication reminders, answering pre-operative questions, and onboarding patients.

To evaluate its AI's “bedside manner,” Hippocratic developed a benchmark to test the model for empathy, and it reportedly scored the highest across all categories of models tested, including GPT-4.

Italy Prepares Job Losses Due to AI: Italy recently allocated 30 million euros ($33 million) to help its citizens improve their digital skills, particularly for jobs at risk of being replaced by automation and technological advances.

Several jobs in Italy could be at risk, including those in transportation and logistics, office and administrative support, production, services, and sales. The program will invest $10 million to train people whose jobs are most at risk of being replaced, and $20 million to help unemployed or inactive people develop digital skills.

Additional Links for “Company Announcements and News Throughout the Industry”:

• Product Launch News Feed

• Venture Capital and Startups News Feed

Thanks for reading this issue of the AstroFeather newsletter!

I’m always looking for ways to improve and would love to hear your constructive feedback about the format and content of the newsletter. You can reply to this email, and I’ll be sure to respond.

See you in the next issue!

If you enjoy AstroFeather weekly content:

1. Share this newsletter

2. Follow the AstroFeather page on LinkedIn

Adides Williams, Founder @ AstroFeather (astrofeather.com)