I’ve spent months talking with founders, investors, and scientists, trying to understand what this technology is and who the players are. Today, I’m going to share my findings. I’ll cover:
What an AI agent is
The major players
The technical bets
The future
Agentic workflows are loops—they can run many times in a row without needing a human involved for each step in the task. A language model will make a plan based on your prompt, utilize tools like a web browser to execute on that plan, ask itself if that answer is right, and close the loop by getting back to you with that answer.
But agentic workflows are an architecture, not a product. It gets even more complicated when you incorporate agents into products that customers will buy.
…
Early reports of GPT-5 are that it is “materially better” and is being explicitly prepared for the use case of AI agents.
As the FlexOS research study “Generative AI at Work” concluded based on a survey amongst knowledge workers, ChatGPT reigns supreme. … 2. AI Tool Usage is Way Higher Than People Expect – Beating Netflix, Pinterest, Twitch. As measured by data analysis platform Similarweb based on global web traffic tracking, the AI tools in this list generate over 3 billion monthly visits.
With 1.67 billion visits, ChatGPT represents over half of this traffic and is already bigger than Netflix, Microsoft, Pinterest, Twitch, and The New York Times.
Something unusual is happening in America. Demand for electricity, which has stayed largely flat for two decades, has begun to surge.
Over the past year, electric utilities have nearly doubled their forecasts of how much additional power they’ll need by 2028 as they confront an unexpected explosion in the number of data centers, an abrupt resurgence in manufacturing driven by new federal laws, and millions of electric vehicles being plugged in.
The tumult could seem like a distraction from the startup’s seemingly unending march toward AI advancement. But the tension, and the latest debate with Musk, illuminates a central question for OpenAI, along with the tech world at large as it’s increasingly consumed by artificial intelligence: Just how open should an AI company be?
…
The meaning of the word “open” in “OpenAI” seems to be a particular sticking point for both sides — something that you might think sounds, on the surface, pretty clear. But actual definitions are both complex and controversial.
In partnership with the National Cancer Institute, or NCI, researchers from the Department of Energy’s Oak Ridge National Laboratory and Louisiana State University developed a long-sequenced AI transformer capable of processing millions of pathology reports to provide experts researching cancer diagnoses and management with exponentially more accurate information on cancer reporting.
The early vibrations of AI have already been shaking the newsroom. One downside of the new technology surfaced at CNET and Sports Illustrated, where editors let AI run amok with disastrous results. Elsewhere in news media, AI is already writing headlines, managing paywalls to increase subscriptions, performing transcriptions, turning stories in audio feeds, discovering emerging stories, fact checking, copy editing and more.
Felix M. Simon, a doctoral candidate at Oxford, recently published a white paper about AI’s journalistic future that eclipses many early studies. Swinging a bat from a crouch that is neither doomer nor Utopian, Simon heralds both the downsides and promise of AI’s introduction into the newsroom and the publisher’s suite.
Unlike earlier technological revolutions, AI is poised to change the business at every level. It will become — if it already isn’t — the beginning of most story assignments and will become, for some, the new assignment editor. Used effectively, it promises to make news more accurate and timely. Used frivolously, it will spawn an ocean of spam. Wherever the production and distribution of news can be automated or made “smarter,” AI will surely step up. But the future has not yet been written, Simon counsels. AI in the newsroom will be only as bad or good as its developers and users make it.
We proposed EMO, an expressive audio-driven portrait-video generation framework. Input a single reference image and the vocal audio, e.g. talking and singing, our method can generate vocal avatar videos with expressive facial expressions, and various head poses, meanwhile, we can generate videos with any duration depending on the length of input video.
New experimental work from Adobe Research is set to change how people create and edit custom audio and music. An early-stage generative AI music generation and editing tool, Project Music GenAI Control allows creators to generate music from text prompts, and then have fine-grained control to edit that audio for their precise needs.
“With Project Music GenAI Control, generative AI becomes your co-creator. It helps people craft music for their projects, whether they’re broadcasters, or podcasters, or anyone else who needs audio that’s just the right mood, tone, and length,” says Nicholas Bryan, Senior Research Scientist at Adobe Research and one of the creators of the technologies.
There’s a lot going on in the world of generative AI, but maybe the biggest is the increasing number of copyright lawsuits being filed against AI companies like OpenAI and Stability AI. So for this episode, we brought on Verge features editor Sarah Jeong, who’s a former lawyer just like me, and we’re going to talk about those cases and the main defense the AI companies are relying on in those copyright cases: an idea called fair use.
The FCC’s war on robocalls has gained a new weapon in its arsenal with the declaration of AI-generated voices as “artificial” and therefore definitely against the law when used in automated calling scams. It may not stop the flood of fake Joe Bidens that will almost certainly trouble our phones this election season, but it won’t hurt, either.
The new rule, contemplated for months and telegraphed last week, isn’t actually a new rule — the FCC can’t just invent them with no due process. Robocalls are just a new term for something largely already prohibited under the Telephone Consumer Protection Act: artificial and pre-recorded messages being sent out willy-nilly to every number in the phone book (something that still existed when they drafted the law).
EIEIO…Chips Ahoy!— from dashmedia.co by Michael Moe, Brent Peus, and Owen Ritz
Here Come the AI Worms — from wired.com by Matt Burgess Security researchers created an AI worm in a test environment that can automatically spread between generative AI agents—potentially stealing data and sending spam emails along the way.
Now, in a demonstration of the risks of connected, autonomous AI ecosystems, a group of researchers have created one of what they claim are the first generative AI worms—which can spread from one system to another, potentially stealing data or deploying malware in the process. “It basically means that now you have the ability to conduct or to perform a new kind of cyberattack that hasn’t been seen before,” says Ben Nassi, a Cornell Tech researcher behind the research.
Last month, OpenAI chief executive Sam Altman finally admitted what researchers have been saying for years — that the artificial intelligence (AI) industry is heading for an energy crisis. It’s an unusual admission. At the World Economic Forum’s annual meeting in Davos, Switzerland, Altman warned that the next wave of generative AI systems will consume vastly more power than expected, and that energy systems will struggle to cope. “There’s no way to get there without a breakthrough,” he said.
I’m glad he said it. I’ve seen consistent downplaying and denial about the AI industry’s environmental costs since I started publishing about them in 2018. Altman’s admission has got researchers, regulators and industry titans talking about the environmental impact of generative AI.
Yesterday, Nvidia reported $22.1 billion in revenue for its fourth fiscal quarter of fiscal 2024 (ending January 31, 2024), easily topping Wall Street’s expectations. The revenues grew 265% from a year ago, thanks to the explosive growth of generative AI.
…
He also repeated a notion about “sovereign AI.” This means that countries are protecting the data of their users and companies are protecting data of employees through “sovereign AI,” where the large-language models are contained within the borders of the country or the company for safety purposes.
??BREAKING: Adobe has created a new 50-person AI research org called CAVA (Co-Creation for Audio, Video, & Animation).
I can’t help but wonder if OpenAI’s Sora has been a wake up call for Adobe to formalize and accelerate their video and multimodal creation efforts?
Nvidia is building a new type of data centre called AI factory. Every company—biotech, self-driving, manufacturing, etc will need an AI factory.
Jensen is looking forward to foundational robotics and state space models. According to him, foundational robotics could have a breakthrough next year.
The crunch for Nvidia GPUs is here to stay. It won’t be able to catch up on supply this year. Probably not next year too.
A new generation of GPUs called Blackwell is coming out, and the performance of Blackwell is off the charts.
Nvidia’s business is now roughly 70% inference and 30% training, meaning AI is getting into users’ hands.
Text to video via OpenAI’s Sora. (I had taken this screenshot on the 15th, but am posting it now.)
We’re teaching AI to understand and simulate the physical world in motion, with the goal of training models that help people solve problems that require real-world interaction.
Introducing Sora, our text-to-video model. Sora can generate videos up to a minute long while maintaining visual quality and adherence to the user’s prompt.
At the University of Pennsylvania, undergraduate students in its school of engineering will soon be able to study for a bachelor of science degree in artificial intelligence.
What can one do with an AI degree? The University of Pennsylvania says students will be able to apply the skills they learn in school to build responsible AI tools, develop materials for emerging chips and hardware, and create AI-driven breakthroughs in healthcare through new antibiotics, among other things.
Google on Monday announced plans to help train people in Europe with skills in artificial intelligence, the latest tech giant to invest in preparing workers and economies amid the disruption brought on by technologies they are racing to develop.
The acceleration of AI deployments has gotten so absurdly out of hand that a draft post I started a week ago about a new development is now out of date.
… The Pace is Out of Control
A mere week since Ultra 1.0’s announcement, Google has now introduced us to Ultra 1.5, a model they are clearly positioning to be the leader in the field. Here is the full technical report for Gemini Ultra 1.5, and what it can do is stunning.
[St. Louis, MO, February 14, 2024] – In a bold move that counters the conventions of more traditional schools, Maryville University has unveiled a substantial $21 million multi-year investment in artificial intelligence (AI) and cutting-edge technologies. This groundbreaking initiative is set to transform the higher education experience to be powered by the latest technology to support student success and a five-star experience for thousands of students both on-campus and online.
From DSC: This would be huge for all of our learning ecosystems, as the learning agents could remember where a particular student or employee is at in terms of their learning curve for a particular topic.
OpenAI on Thursday announced its first partnership with a higher education institution.
Starting in February, Arizona State University will have full access to ChatGPT Enterprise and plans to use it for coursework, tutoring, research and more.
The partnership has been in the works for at least six months.
ASU plans to build a personalized AI tutor for students, allow students to create AI avatars for study help and broaden the university’s prompt engineering course.
The collaboration between ASU and OpenAI brings the advanced capabilities of ChatGPT Enterprise into higher education, setting a new precedent for how universities enhance learning, creativity and student outcomes.
“ASU recognizes that augmented and artificial intelligence systems are here to stay, and we are optimistic about their ability to become incredible tools that help students to learn, learn more quickly and understand subjects more thoroughly,” ASU President Michael M. Crow said. “Our collaboration with OpenAI reflects our philosophy and our commitment to participating directly to the responsible evolution of AI learning technologies.”
AI <> Academia — from drphilippahardman.substack.com by Dr. Philippa Hardman What might emerge from ASU’s pioneering partnership with OpenAI?
Phil’s Wish List #2: Smart Curriculum Development ChatGPT assists in creating and updating course curricula, based on both student data and emerging domain and pedagogical research on the topic.
Output: using AI it will be possible to review course content and make data-informed automate recommendations based on latest pedagogical and domain-specific research
Potential Impact: increased dynamism and relevance in course content and reduced administrative lift for academics.
Unlike traditional leadership, adaptable leadership is not bound by rigid rules and protocols. Instead, it thrives on flexibility. Adaptable leaders are willing to experiment, make course corrections, and pivot when necessary. Adaptable leadership is about flexibility, resilience and a willingness to embrace change. It embodies several key principles that redefine the role of leaders in organizations:
Embracing uncertainty
Adaptable leaders understand that uncertainty is the new norm. They do not shy away from ambiguity but instead, see it as an opportunity for growth and innovation. They encourage a culture of experimentation and learning from failure.
Empowering teams
Instead of dictating every move, adaptable leaders empower their teams to take ownership of their work. They foster an environment of trust and collaboration, enabling individuals to contribute their unique perspectives and skills.
Continuous learning
Adaptable leaders are lifelong learners. They are constantly seeking new knowledge, stay informed about industry trends and encourage their teams to do the same. They understand that knowledge is a dynamic asset that must be constantly updated.
Major AI in Education Related Developments this week — from stefanbauschard.substack.com by Stefan Bauschard ASU integrates with ChatGPT, K-12 AI integrations, Agents & the Rabbit, Uruguay, Meta and AGI, Rethinking curriculum
“The greatest risk is leaving school curriculum unchanged when the entire world is changing.” Hadi Partovi, founder Code.org, Angel investor in Facebook, DropBox, AirBnb, Uber
Tutorbots in college. On a more limited scale, Georgia State University, Morgan State University, and the University of Central Florida are piloting a project using chatbots to support students in foundational math and English courses.
Pioneering AI-Driven Instructional Design in Small College Settings — from campustechnology.com by Gopu Kiron For institutions that lack the budget or staff expertise to utilize instructional design principles in online course development, generative AI may offer a way forward.
Unfortunately, smaller colleges — arguably the institutions whose students are likely to benefit the most from ID enhancements — frequently find themselves excluded from authentically engaging in the ID arena due to tight budgets, limited faculty online course design expertise, and the lack of ID-specific staff roles. Despite this, recent developments in generative AI may offer these institutions a low-cost, tactical avenue to compete with more established players.
There’s a new AI from Google DeepMind called AlphaGeometry that totally nails solving super hard geometry problems. We’re talking problems so tough only math geniuses who compete in the International Mathematical Olympiad can figure them out.
Introducing the GPT Store — from OpenAI We’re launching the GPT Store to help you find useful and popular custom versions of ChatGPT.
It’s been two months since we announced GPTs, and users have already created over 3 million custom versions of ChatGPT. Many builders have shared their GPTs for others to use. Today, we’re starting to roll out the GPT Store to ChatGPT Plus, Team and Enterprise users so you can find useful and popular GPTs. Visit chat.openai.com/gpts to explore.
Introducing ChatGPT Team — from openai.com We’re launching a new ChatGPT plan for teams of all sizes, which provides a secure, collaborative workspace to get the most out of ChatGPT at work.
ChatGPT Team offers access to our advanced models like GPT-4 and DALL·E 3, and tools like Advanced Data Analysis. It additionally includes a dedicated collaborative workspace for your team and admin tools for team management. As with ChatGPT Enterprise, you own and control your business data—we do not train on your business data or conversations, and our models don’t learn from your usage. More details on our data privacy practices can be found on our privacy page and Trust Portal.
GPT Store — from theneurondaily.com by Noah Edelman & Pete Huang
The App Store for ChatGPTs is here.
OpenAI finally launched its GPT Store—a hub offering access to over 3 million GPTs, for paid users (#sorrynotsorry).
If you missed pt. 1, pt. 2, and pt. 3 of our GPTs analysis, here’s the TLDR: GPTs are customized versions of ChatGPT pre-loaded with prompts or context, each designed to be good at specific tasks.
There’s a GPT for everything, like one for lesson plans, one that crunches numbers, and one that recommends books you’ll buy but never read.
How to stand out on the GPT Store The low barrier to entry for making GPTs will make earning money on the GPT store difficult. Not everyone will make tons of money off their GPT, but I think those with more chances of success will:
Use custom actions: This is a feature that allows your GPT to connect to an API. Connecting to APIs gives your GPT new functionalities that others won’t be able to replicate unless they have access to the API (here you can see my tutorial on how to add custom action to your GPT)
Use knowledge: Knowledge is a feature that allows you to add files to your GPT. Adding exclusive information could enrich your GPT and help it stand out from the pack. Just remember that files can be downloaded when the code interpreter is enabled.
App stores are ginormous businesses. According to CNBC’s estimates, Apple’s App Store grossed north of $70 Billion in 2022. That’s more revenue than Spotify, Shopify and Airbnb generated in the same year — combined.
When you look at the size of the opportunity that app stores built on top of popular platforms unlock, OpenAI’s latest move to launch a GPT Store is another bold bet by the startup that’s already leading the LLM and chatbot markets with GPT-4 and ChatGPT.
Announced [on 1/10/24], the GPT Store is a place for ChatGPT users to find custom versions of the chatbot that are designed for specific use cases.
OpenAI plans to launch a store for GPTs, custom apps based on its text-generating AI models (e.g. GPT-4), sometime in the coming week.
…
The GPT Store was announced last year during OpenAI’s first annual developer conference, DevDay, but delayed in December — almost certainly due to the leadership shakeup that occurred in November, just after the initial announcement.
verb
(of artificial intelligence) to produce false information contrary to the intent of the user and present it as if true and factual. Example: When chatbots hallucinate, the result is often not just inaccurate but completely fabricated.
Soon, every employee will be both AI builder and AI consumer— from zdnet.com by Joe McKendrick, via Robert Gibson on LinkedIn “Standardized tools and platforms as well as advanced low- or no-code tech may enable all employees to become low-level engineers,” suggests a recent report.
The time could be ripe for a blurring of the lines between developers and end-users, a recent report out of Deloitte suggests. It makes more business sense to focus on bringing in citizen developers for ground-level programming, versus seeking superstar software engineers, the report’s authors argue, or — as they put it — “instead of transforming from a 1x to a 10x engineer, employees outside the tech division could be going from zero to one.”
Along these lines, see:
TECH TRENDS 2024 — from deloitte.com Six emerging technology trends demonstrate that in an age of generative machines, it’s more important than ever for organizations to maintain an integrated business strategy, a solid technology foundation, and a creative workforce.
The ruling follows a similar decision denying patent registrations naming AI as creators.
The UK Supreme Court ruled that AI cannot get patents, declaring it cannot be named as an inventor of new products because the law considers only humans or companies to be creators.
The New York Times sued OpenAI and Microsoft for copyright infringement on Wednesday, opening a new front in the increasingly intense legal battle over the unauthorized use of published work to train artificial intelligence technologies.
…
The suit does not include an exact monetary demand. But it says the defendants should be held responsible for “billions of dollars in statutory and actual damages” related to the “unlawful copying and use of The Times’s uniquely valuable works.” It also calls for the companies to destroy any chatbot models and training data that use copyrighted material from The Times.
On this same topic, also see:
? The historic NYT v. @OpenAI lawsuit filed this morning, as broken down by me, an IP and AI lawyer, general counsel, and longtime tech person and enthusiast.
Tl;dr – It’s the best case yet alleging that generative AI is copyright infringement. Thread. ? pic.twitter.com/Zqbv3ekLWt
ChatGPT and Other Chatbots
The arrival of ChatGPT sparked tons of new AI tools and changed the way we thought about using a chatbot in our daily lives.
Chatbots like ChatGPT, Perplexity, Claude, and Bing Chat can help content creators by quickly generating ideas, outlines, drafts, and full pieces of content, allowing creators to produce more high-quality content in less time.
These AI tools boost efficiency and creativity in content production across formats like blog posts, social captions, newsletters, and more.
Microsoft is getting ready to upgrade its Surface lineup with new AI-enabled features, according to a report from Windows Central. Unnamed sources told the outlet the upcoming Surface Pro 10 and Surface Laptop 6 will come with a next-gen neural processing unit (NPU), along with Intel and Arm-based options.
With the AI-assisted reporter churning out bread and butter content, other reporters in the newsroom are freed up to go to court, meet a councillor for a coffee or attend a village fete, says the Worcester News editor, Stephanie Preece.
“AI can’t be at the scene of a crash, in court, in a council meeting, it can’t visit a grieving family or look somebody in the eye and tell that they’re lying. All it does is free up the reporters to do more of that,” she says. “Instead of shying away from it, or being scared of it, we are saying AI is here to stay – so how can we harness it?”
This year, I watched AI change the world in real time.
From what happened, I have no doubts that the coming years will be the most transformative period in the history of humankind.
Here’s the full timeline of AI in 2023 (January-December):
What to Expect in AI in 2024 — from hai.stanford.edu by Seven Stanford HAI faculty and fellows predict the biggest stories for next year in artificial intelligence.
.webp)
