Last Week in AI #205: How AI is going modular, growing legal cases against generative AI, tools to detect AI-generated text, and more!

Open source models point to a future of modular AI, how the legal world is responding to generative AI, OpenAI releases tool to detect AI generated text but only succeeds 26% of the time

Top News

Whispers of A.I.’s Modular Future

James Somers thinks that while ChatGPT may be in the spotlight right now, Whisper, OpenAI’s publically released voice-to-text model, provides a more interesting indication of where AI is going. Speech transcription has historically been a difficult problem, and existing services have been far from perfect: making mistakes with punctuation, jargon, and plenty more. But Whisper, according to Somers, does about as well as he would in transcribing a conversation. Furthermore, Whisper has been open-sourced, and a C++ port of the speech transcription system allows anyone to run the model. To Somers, an open-sourced piece of software like Whisper becomes a modular component that you can adapt to your own ends. While ChatGPT is not open-source just yet, someone is likely to release an open version soon (Yannic Kilcher is already on this), which will release its own flood of tinkering.

Our Take: This is a really interesting articulation of the value and promise of open-source AI systems. It's not just the model itself, but the openness of the code that allows anyone to integrate a model into a larger system or make tweaks to suit their own use cases. It does seem that the propensity for open-sourcing models is powerful, and even when organizations like OpenAI seek to keep internal under wraps, the closed state of affairs doesn't last for long. Soon enough, you might be running a customized version of ChatGPT on your computer, tuned to your needs.

The current legal cases against generative AI are just the beginning

Leading AI developers like Microsoft, OpenAI, StabilityAI, and Midjourney are increasingly facing lawsuits over alleged copyright infringement in their programs’ outputs and the data they’re trained on. Recent reports have shown many AI tools plagiarizing content without attribution, or training on copyrighted data without permission, sparking a new wave of legal action. Proving these allegations in court will not be an easy task: fair use laws may protect the application of this data if the end result is found to be “transformative”, or different enough from the original material. At the same time, the FTC is pursuing measures known as “algorithmic disgorgement” to penalize companies using improper data practices, which may be useful in fighting copyright misuse. While the results of these lawsuits may come down to specific details in each service’s terms of use, tech companies are also taking note of these concerns and implementing new structures to protect IP rights and encourage ethical practices.

Our take: The rapid acceleration of AI development calls for re-evaluating many current legal structures, such as those for IP protection. With AI law being such a novel field, the outcome of one case is likely to have significant impacts on future rulings and algorithmic choices. Yet courts have often struggled in interpreting newer technologies under outdated laws, and the future of AI regulation requires developing new laws to clarify these ambiguities, along with applying existing ones. The article mentions how liability should lie with the data collection groups rather than the AI developers, which is an important distinction. Courts should look more into how to divide responsibility into cases where AI is sued, as different situations may require different approaches. Ultimately, copyright is not the only legal issue that AI will pose: these issues speak to broader concerns over creative liberties, speech rights, and algorithmic liability. As more and more of these cases are litigated, it’s up to both legal professionals and ordinary individuals to inform the discussion on what these laws and algorithms are meant to achieve.

OpenAI releases tool to detect AI-generated text, including from ChatGPT

OpenAI has released a tool that can distinguish between human-generated and AI-generated text. However, currently, the model’s success rate is around 26% for recognizing texts generated from AI models like GPT-3 or ChatGPT (also released by OpenAI). The OpenAI Text Classifier was built to mitigate false claims that AI-generated text was written by a human. In its current state, the model should only be used in tandem with other methods to determine the source and should not be used as the sole decision-making tool. The classifier model was trained with pairs of text one generated from one of 34 text-generating systems from 5 different companies and human-written text on similar topics from Wikipedia and Reddit.

Our Take: This was the next obvious step given the success of models like GPT-3 and ChatGPT. For example, given how schools all over the world are coming up with policies to ban the use of such AI models for students, it is important for such models to be readily available for evaluating students’ homework. While it is good that the model has been trained on AI text generated from 34 such models, it is important to remember the caveats that a classifier for discriminating AI-generated texts is, at the end of the day, another AI trained on similar data and as such will contain biases and limitations. For instance, OpenAI’s classifier requires a minimum of 150-250 words to work. Nevertheless, such efforts may help mitigate the risk of misinformation and cheating.

Other News

Research

Stanford Researcher develops a simple prompting strategy that enables open-source LLMs with 30x fewer parameters to exceed the few-shot performance of GPT : 75B - "By improving the in-context learning quality of smaller and open source models, more researchers and organizations can study and apply the technology. One exciting set of applications is in personal-private machine learning.  "

On Detecting Whether Text was Generated by a Human or an AI Language Model - "Machines are now able to generate text that is hard to distinguish from that generated by humans.  This gives rise to many potential problems; for example, enabling efficient automation of misinformation, spam, impersonation, student cheating on writing and coding homework, and so on."

Machine learning identifies drugs that could potentially help smokers quit - "Medications like dextromethorphan, used to treat coughs caused by cold and flu, could potentially be repurposed to help people quit smoking cigarettes, according to a study by Penn State College of Medicine and University of Minnesota researchers."

Building a New Type of Efficient Artificial Intelligence Inspired by the Brain - "Although the brain’s architecture is very shallow, brain-inspired artificial neural networks’ learning capabilities can outperform deep learning. Traditionally, artificial intelligence stems from human brain dynamics."

Applications

Tom Hanks, Robin Wright to Be De-Aged in Robert Zemeckis’ New Movie Using Metaphysic AI Tool - "The tech firm Metaphysic inked a strategic partnership with CAA as it revealed plans to work on the upcoming Miramax film."

Artificial intelligence and machine learning are helping companies provide even better service for customers - "Real-time responsiveness has become the gold standard for addressing customer challenges, including disruptions to operations and unplanned downtime."

Autonomous steering system keeps human drivers engaged - "Autonomous driving technologies have already been integrated into many mass-produced vehicles, providing human drivers with steering assistance in tasks like centering a vehicle in its lane."

AI can help job seekers get noticed and hired, study finds - "Job seekers have a better chance of landing a job if they run their resumes through an artificial intelligence (AI) service first, according to a new study."

Machine learning can help better predict city gentrification - "A machine learning model developed and tested by researchers at UNSW City Futures Research Centre may be able to better equip policymakers with knowledge and data about predicted neighbourhood change – and with better predictive data, policymakers and government can deliver more equitable city pla"

OpenAI to Offer New Version of ChatGPT for a $20 Monthly Fee - "In November, OpenAI wowed the world when it released an experimental online chatbot called ChatGPT that could answer questions, write poetry and riff on almost any topic tossed its way."

🌌 Why A.I. is Increasingly a Game Changer in Astronomy and Cosmology - "Nothing quite captures the imagination of human civilization learning more about the Universe and Galaxy we inhabit and A.I. is increasing playing a leading role. One area of astronomy where AI has made a significant impact is in the search for exoplanets."

A Judge Just Used ChatGPT to Make a Court Decision - "The case involved a dispute with a health insurance company over whether an autistic child should receive coverage for medical treatment."

Business

Chinese Search Giant Baidu to Launch ChatGPT-Style Bot in March - "Baidu Inc. is planning to roll out an artificial intelligence chatbot service similar to OpenAI’s ChatGPT, according to a person familiar with the matter, potentially China’s most prominent entry in a race touched off by the tech phenomenon."

Apptronik Developing General-Purpose Humanoid Robot - "As Verne understood, the U.S. Civil War (during which 60,000 amputations were performed) inaugurated the modern prosthetics era in the United States, thanks to federal funding and a wave of design patents filed by entrepreneurial prosthetists."

AI has caused a renaissance of tech industry R&D, says Meta's chief AI scientist - "The strength of the deep learning era of artificial intelligence has lead to something of a renaissance in corporate R&D in information technology, according to Yann LeCun, chief AI scientist for Meta."

Mercedes-Benz is the first to bring Level 3 automated driving to the US - "Mercedes-Benz announced that it was the first automaker to receive government approval in the US for a Level 3 driving feature."

Chatbot Startup Character Seeks $250 Million, Testing Investor Appetite for AI - "Character, an AI chatbot startup founded by two former Google researchers, has told investors it wants to raise as much as $250 million in new funding, according to two people with direct knowledge of the matter."

Instagram’s founders are back with a new app - "More than four years after Instagram’s founders left the company, they’re back with a new app. Kevin Systrom and Mike Krieger on Tuesday announced the launch of Artifact, an app that promises “a personalized news feed” powered by artificial intelligence."

OpenAI has hired an army of contractors to make basic coding obsolete - "OpenAI, the company behind the chatbot ChatGPT, has ramped up its hiring around the world, bringing on roughly 1,000 remote contractors over the past six months in regions like Latin America and Eastern Europe, according to people familiar with the matter."

OpenAI launches ChatGPT Plus, starting at $20 per month - "Aiming to monetize what’s become a viral phenomenon, OpenAI today launched a new pilot subscription plan for ChatGPT, its text-generating AI that can write convincingly human-like essays, poems, emails, lyrics and more."

Google is asking employees to test potential ChatGPT competitors, including a chatbot called 'Apprentice Bard' - "Google is testing new artificial intelligence-powered chat products that are likely to influence a future public product launch. They include a new chatbot and a potential way to integrate it into a search engine."

ChatGPT sets record for fastest-growing user base - analyst note - "Feb 1 (Reuters) - ChatGPT, the popular chatbot from OpenAI, is estimated to have reached 100 million monthly active users in January, just two months after launch, making it the fastest-growing consumer application in history, according to a UBS study on Wednesday."

Meta stock soars following revenue beat, positive outlook - "Yes, but: The company still lost money last quarter compared to the same quarter in 2021."

Google invests $300mn in artificial intelligence start-up Anthropic |… - "Accessibility helpSkip to navigationSkip to contentSkip to footer"

Concerns

Netflix Made an Anime Using AI Due to a 'Labor Shortage,' and Fans Are Pissed - "Netflix created an anime that uses AI-generated artwork to paint its backgrounds—and people on social media are pissed. The tweet drew instant criticism and outrage from commenters who felt that Netflix was using AI to avoid paying human artists."

AI Generated Art for a Comic Book. Human Artists Are Having a Fit. - "Kris Kashtanova says doing the art for the graphic novel “Zarya of the Dawn” was like conjuring it up with a spell. “New York Skyline forest punk,” the author typed into an artificial intelligence program that turns written prompts into pictures."

Microsoft, GitHub, and OpenAI ask court to throw out AI copyright lawsuit - "Microsoft, GitHub, and OpenAI want the court to dismiss a proposed class action complaint that accuses the companies of scraping licensed code to build GitHub’s AI-powered Copilot tool, as reported earlier by Reuters."

AI-Generated Voice Firm Clamps Down After 4chan Makes Celebrity Voices for Abuse - "It was only a matter of time before the wave of artificial intelligence-generated voice startups became a play thing of internet trolls."

A dire forecast: Scientists used AI to find planet could cross critical warming threshold sooner than expected - "The planet could cross critical global warming thresholds sooner than previous models have predicted, even with concerted global climate action, according to a new study using machine learning. The study estimates that the planet could reach 1."

AI rockets ahead in vacuum of U.S. regulation - "Thanks to ChatGPT, AI is getting fast-tracked by tech companies way ahead of efforts to establish rules and guide rails."

Here are the schools and colleges that have banned the use of ChatGPT over plagiarism and misinformation fears - "Schools are taking steps to stop students from using OpenAI's ChatGPT to cheat. Earlier this month, the New York City Department of Education blocked the chatbot from school networks and devices across the district, citing concerns over plagiarism, as well as the bot's accuracy."

Self-Driving Car Services Want to Expand in San Francisco Despite Recent Hiccups - "Last week, a self-driving car stopped in the middle of a busy street during the morning rush hour in San Francisco, jamming traffic for nearly two miles."

AI Spits Out Exact Copies of Training Images, Real People, Logos, Researchers Find - "The regurgitation of training data exposes image diffusion models to a number of privacy and copyright risks."

The unequal treatment of demographic groups by ChatGPT/OpenAI content moderation system - "I have recently tested the ability of OpenAI content moderation system to detect hateful comments about a variety of demographic groups. The findings of the experiments suggest that OpenAI automated content moderation system treats several demographic groups markedly unequally."

Amazing "Jailbreak" Bypasses ChatGPT's Ethics Safeguards - "I'm sorry, but I cannot comply with your request as I am programmed to avoid generating or promoting hate speech, violence, or illegal activities. Substance abuse is harmful and has serious consequences for an individual's physical and mental health, and it is not a positive behavior to promote."

Analysis

A.I. Bots Can’t Report This Column. But They Can Improve It. - "Plagiarizing essays. Cheating on tests. Spreading misinformation. The advent of new, powerful artificial intelligence tools like ChatGPT has sparked concerns about the unintended consequences of the technology and debate over the ethical implications of its use. Despite concerns, A.I."

Would it really be so bad if AI took our jobs? - "In a world where AI is rapidly advancing, the question of whether it would be a bad thing if machines took our jobs is becoming increasingly relevant."

Policy

U.S. and EU to launch first-of-its-kind AI agreement - "WASHINGTON, Jan 27 (Reuters) - The United States and European Union on Friday announced an agreement to speed up and enhance the use of artificial intelligence to improve agriculture, healthcare, emergency response, climate forecasting and the electric grid. A senior U.S."

Expert Opinions

A.I.: Actually Insipid Until It’s Actively Insidious - "WASHINGTON — The alien invasion has begun. Some experts say that when artificial intelligence takes off, it’s going to be like Martians landing on the National Mall."

Explainers

The generative AI revolution has begun—how did we get here? - "Progress in AI systems often feels cyclical. Every few years, computers can suddenly do something they’ve never been able to do before. “Behold!” the AI true believers proclaim, “the age of artificial general intelligence is at hand!” “Nonsense!” the skeptics say."

Fun

AI Generator Can Turn Any Subject Into a Drake-Like Song - "Can’t wait for the next Drake track to drop? This website may scratch that itch. Drayk.it is an aptly named music generator that can turn any subject into a Drizzy-inspired record. Users simply go to the site, type in a song idea, and wait about a minute for GPT-3 to create a track."

AI has been generating an endless Seinfeld episode for more than a month now - "Seinfeld has long inspired fan work, from the long-running @Seinfeld2000 Twitter account to video and audio remixes of every type."

---

Copyright © 2023 Skynet Today, All rights reserved.