Last Week in AI #199: Economic (dis)incentives of ChatGPT, TikTok to be more public about its algorithm, Google's response to AI chatbots, and more!
What will happen to services like ChatGPT when they need to make money, TikTok to reveal more about its algorithm, Google issues "code red" over ChatGPT
Generative AI has the potential to transform various industries, from healthcare to art and education. Yet the seemingly unlimited magic of systems like ChatGPT comes at a high cost (up to $3 million a month) – which can’t be sustained forever. What will happen when these technologies are forced to turn a profit? The answer may be found by observing the path of previous advances – whose promises quickly fell to profit-seeking interests. Many predicted the internet would bring a new democratic future until it became controlled by major companies determined to keep users tied to their platforms. New technologies usually disrupt the most vulnerable industries, and small businesses may be displaced much faster than major ones.
While ChatGPT may lose its free status eventually, new initiatives are sprouting up to help these technologies stay freely accessible. Petals, an open-source community project whose code was released last month, hopes to reduce the cost of running massive AI systems by creating a distributed network of volunteer users. While the public nature of the project leaves it open to many security threats, it represents a new attempt to democratize generative AI and lower the barrier to using it.
Our Take: Technological development doesn’t happen in a vacuum. The trajectory that AI follows will inherently be affected by the economic, political, and social context surrounding it, as shown by its unequal distribution of benefits and misuse by authoritarian actors. Technologies may bring in profits for those deploying them while keeping costs artificially high: as the article explains, “the cost of health care in America is not directly driven by the salary of medical doctors.” The Petals system highlights the importance of thinking critically about the future of AI: if we want it to bring the most public benefit, we have to create the structures for it to do so. Otherwise, the surplus value created by these systems is likely to quietly reinforce existing structures of power, even if the companies at the top of those structures change.
Recommendation algorithms play a commanding role in our online lives, and we know little about how they work from the outside. TikTok's algorithm, in particular, is known to make the video-sharing app uncannily addictive. This last Tuesday, TikTok announced a move towards greater transparency: a new "Why you're seeing this video" feature will tell users why TikTok's algorithm recommends certain videos in their feeds. In the midst of controversy and TikTok's being banned from government devices, it's unsurprising that TikTok would make attempts to alleviate continued scrutiny and ease concerns. Indeed, TikTok has also added tools for users to customize content recommendations and parental controls.
Our Take: I don't use TikTok and have no plans to, so anything I say about this will be from someone who hasn't had direct experience with the algorithm/UI. I think it's nice that TikTok has made changes to increase transparency and allow users greater control over what they see--but inevitably these changes are driven by incentives that ultimately boil down to what's best for TikTok. The algorithmic transparency decision is interesting--it's hard to know from the outside how representative the information users will see is of what TikTok's algorithm actually does. Certainly, it will target content at people based on creators they follow and geographic locations; TikTok may even be willing to release more granular information. But how far are they willing to go, and what's missing? My skepticism is founded on the basis that such an initiative is likely to be in service of ensuring regulators and users don't further scrutinize TikTok and its algorithms. Allowing people to think they know what's going on behind the scenes is a very effective way to evade and distract from real concerns.
This article discusses the potential impact of chatbots, specifically ChatGPT, on the internet search industry. ChatGPT, developed by OpenAI, is a large language model that can provide information in clear, simple sentences and even generate ideas from scratch. Some believe that the technology behind ChatGPT could pose a threat to traditional search engines like Google, which has relied on its search engine as a primary gateway to the internet for more than 20 years. However, there are concerns about the accuracy and bias of information provided by language models, as well as the impact on Google's revenue from digital ads, which accounted for over 80% of the company's revenue in the previous year. It is also noted that Google has been working on chatbots for several years and has its own chatbot technology, called LaMDA, but Google’s models are not publicly accessible. Google does use its language models in more subtle ways to refine search results, and this is likely the approach the company will take in the near future.
Our take: I think it’s premature to say that ChatGPT-like services will pose an imminent threat to Google’s core ads business. Inaccuracy and bias aside, the economics of deploying these models is a real challenge, and the types of searches language models excel at may not be the ones that are actually monetizable. It is clear that the prevalence and accessibility of large language models will change how many tech services are done, and it’ll be interesting to see how large incumbents react in the coming months.
OpenAI releases Point-E, an AI that generates 3D models - "The next breakthrough to take the AI world by storm might be 3D model generators. This week, OpenAI open-sourced Point-E, a machine learning system that creates a 3D object given a text prompt."
PubMedGPT 2.7B - "We introduce a new 2.7B parameter language model trained on biomedical literature which delivers an improved state of the art for medical question answering."
Study shows how machine learning could predict rare disastrous events, like earthquakes or pandemics - "Researchers from Brown and MIT suggest how scientists can circumvent the need for massive data sets to forecast extreme events with the combination of an advanced machine learning system and sequential sampling techniques"
New software based on Artificial Intelligence helps to interpret complex data - "Experimental data is often not only highly dimensional, but also noisy and full of artefacts. This makes it difficult to interpret the data."
AI Model Can Help Detect Collapsed Lung Using Chest X-rays - "The artificial intelligence model accurately detected pneumothorax, or a collapsed lung, and exceeded FDA guidelines for computer-assisted triage devices."
Petals is creating a free, distributed network for running text-generating AI - "BigScience, a community project backed by startup Hugging Face with the goal of making text-generating AI widely available, is developing a system called Petals that can run AI like ChatGPT by joining resources from people across the internet."
Words prove their worth as teaching tools for robots - "Exploring a new way to teach robots, Princeton researchers have found that human-language descriptions of tools can accelerate the learning of a simulated robotic arm lifting and using a variety of tools."
2022: A Year Full of Amazing AI papers- A Review [WORK IN PROGRESS] 📌 - "A curated list of the latest breakthroughs in AI by release date with a clear video explanation, link to a more in-depth article, and code."
This AI Paper Presents PaletteNeRF, a Novel Method for Photorealistic Appearance Editing of Neural Radiance Fields (NeRF) Based on 3D Color Decomposition - "The capacity of Neural Radiance Fields (NeRF) and its derivatives to accurately recreate real-world 3D scenes from 2D photos and allow high-quality, photorealistic new view synthesis has garnered more and more interest in recent years."
Artificial Intelligence helped connect a Holocaust survivor with photos of her past - "NPR's Juana Summers talks with software engineer Daniel Patt about his website "From Numbers to Names," which uses artificial intelligence to find photos of victims and survivors of the Holocaust."
2 coders made a teleprompter AI that suggests smart things to say during your work meetings - "Two coders have shared a demo of a project they're calling Tele-Prompt — an AI-bot that listens in on work meetings and simultaneously generates interesting quotes that you could choose to interject into the conversation."
Modeling Key World Cup Moments with Machine Learning - "For The Times’s analysis of the 2022 FIFA World Cup, R&D and The New York Times graphics desk collaborated on a series of interactive articles that enable a viewer to see key plays in 3D. To accomplish this, R&D developed a workflow to swiftly create 3D models of athletes in action from a single photograph. "
Krispy Kreme CEO: Robots will start frosting and filling doughnuts 'within the next 18 months’ - "Krispy Kreme (DNUT) is aiming to cut time in its doughnut production line through automation. The Winston-Salem, North Carolina-based company announced the investment as part of its Investor Day, along with its long-term outlook of generating $2.15 billion in revenue by fiscal year 2026."
With the help of machine learning, DC’s Security Bulldog wants to build better cyber defense - "Jeff Majka created the company in 2021 to help cyber professionals reduce the amount of time spent searching for potential problems with their software."
‘South Park’ Creators Trey Parker and Matt Stone Land $20 Million in Funding for Their Deepfake VFX Studio - "Trey Parker and Matt Stone are going deeper into Deep Voodoo. The creators of “South Park” have secured a $20 million investment for their AI entertainment startup Deep Voodoo."
OpenAI Announce a New Embedding Model Which is Significantly More Capable, Cost-Effective, and Simpler to Use - "Embeddings are representations of concepts in the form of sequences of numbers, which facilitates the computer’s ability to grasp the connections between those concepts. "
Quora launches Poe, a way to talk to AI chatbots like ChatGPT - "Signaling its interest in text-generating AI systems like ChatGPT, Quora this week launched a platform called Poe that lets people ask questions, get instant answers and have a back-and-forth dialogue with AI."
Natural Language Processing Market Size to Reach USD 98.05 Billion in 2030 | Emergen Research - "The natural language processing market size reached USD 12.43 Billion in 2021 and is expected to register a CAGR of 25.7% during the forecast period, according to latest analysis by Emergen Research."
How Amazon Uses AI To Automate Work In Its Corporate Headquarters - "Nearly all companies will soon have workplace AI on par with the tech giants. Here's what to expect."
ArtStation is hiding images protesting AI art on the platform - "Art platform ArtStation is removing images protesting AI-generated art from its homepage, claiming that the content violates its Terms of Service. "
ChatGPT Has Infiltrated Twitter Replies - "As Twitter continues to circle the drain fueled by chaos from CEO Elon Musk, one user noticed something peculiar about tweet replies."
‘Full-on robot writing’: the artificial intelligence challenge facing universities - "“Waiting in front of the lecture hall for my next class to start, and beside me two students are discussing which AI program works best for writing their essays. Is this what I’m marking? AI essays?”"
Face Recognition Tech Gets Girl Scout Mom Booted from Rockettes Show — Due to Where She Works - "A recent incident at Radio City Music Hall involving the mother of a Girl Scout is shedding light on the growing controversy of facial recognition, as critics claim it is being used to target perceived enemies — in this case, by one of the most famous companies in the country."
People Are Not Happy That the SF Ballet Used AI-Generated Art To Promote ‘The Nutcracker’ - "Last month, ahead of their annual December production of The Nutcracker, the San Francisco Ballet posted an illustration of the show’s iconic protagonist to its Instagram account. But unlike Drosselmeyer’s familiar wooden soldier, human hands did not create this nutcracker."
Tesla ‘full self-driving’ triggered an eight-car crash, a driver tells police - "A driver told authorities that their Tesla’s “full-self-driving” software braked unexpectedly and triggered an eight-car pileup in the San Francisco Bay Area last month that led to nine people being treated for minor injuries including one juvenile who was hospitalized"
TikTokers are exploiting a dumb AI to generate giant anime boobs - "TikTok is stuffed with all sorts of filters that make the app a little more chaotic. Often they’ll take off and inspire a massive trend. Right now, one such trend involves using the AI Manga filter to generate a portrait of oneself, in anime style."
Kickstarter shut down the campaign for AI porn group Unstable Diffusion amid changing guidelines - "The group trying to monetize AI porn generation, Unstable Diffusion, raised more than $56,000 on Kickstarter from 867 backers. Now, as Kickstarter changes its thinking about what kind of AI-based projects it will allow, the crowdfunding platform has shut down Unstable Diffusion’s campaign."
Cheerful chatbots don't necessarily improve customer service - "Imagine messaging an artificial intelligence (AI) chatbot about a missing package and getting the response that it would be "delighted" to help. Once the bot creates the new order, they say they are "happy" to resolve the issue."
AI-Generated Comic Book Could Lose Copyright Protection - "USPTO ‘overlooked’ the use of text-to-image generator"
Study suggests a robot levy -- but only a modest one -- could help combat the effects of automation on income inequality in the U.S. - "What if the U.S. placed a tax on robots? The concept has been publicly discussed by policy analysts, scholars, and Bill Gates (who favors the notion)."
Portman Introduces Bill to Guard Against Artificial Intelligence Bias - "Stopping Unlawful Negative Machine Impacts through National Evaluation Act to clarify that existing civil rights laws apply to decisions made by AI systems just as if those decisions were made by humans."
California passes law banning Tesla from calling software FSD - "The bill was sponsored after the California Department of Motor Vehicles claimed that Tesla’s FSD was “false advertising.”"
Hayao Miyazaki Calls Artificial Intelligence Animation ‘An Insult To Life Itself’ - "Studio Ghibli director Hayao Miyazaki, responsible for such animation classics as “Spirited Away” and “My Neighbor Totoro,” recently came out of retirement for a new short film entitled “Boro the Caterpillar” and an untitled feature film."
AI's Jurassic Park Moment - "Something incredible is happening in AI right now, and it's not entirely to the good."
The End of Programming - "I believe the conventional idea of "writing a program" is headed for extinction, and indeed, for all but very specialized applications"
A brief history of diffusion, the tech at the heart of modern image-generating AI - "Text-to-image AI exploded this year as technical advances greatly enhanced the fidelity of art that AI systems could create."
Could the future of art lie in AI technology? - BBC News - "Technology is shaking up the art world, with the rise of images being made by artificial intelligence (AI). "
Here's What It Sounds Like When AI Writes Christmas Lyrics - ""Santa's back in town, with a mighty sack. He's ready to fight against the fascists' attack." Those lyrics probably won't be on the lips of many carolers this Christmas, but as far as songs about St. Nick and authoritarianism go, they're pretty catchy. Thanks, AI."
Copyright © 2022 Skynet Today, All rights reserved.