The mounting human and environmental costs of generative AI — from by Sasha Luccioni
Op-ed: Planetary impacts, escalating financial costs, and labor exploitation all factor.
Excerpt:
Over the past few months, the field of artificial intelligence has seen rapid growth, with wave after wave of new models like Dall-E and GPT-4 emerging one after another. Every week brings the promise of new and exciting models, products, and tools. It’s easy to get swept up in the waves of hype, but these shiny capabilities come at a real cost to society and the planet.
Downsides include the environmental toll of mining rare minerals, the human costs of the labor-intensive process of data annotation, and the escalating financial investment required to train AI models as they incorporate more parameters.
Let’s look at the innovations that have fueled recent generations of these models—and raised their associated costs.
Also relevant/see:
ChatGPT is Thirsty: a mini exec briefing — from BrainyActs #040 | ChatGPT’s Thirst Problem
Between the growing number of data centers + the exponential growth in consumer-grade Generative AI tools, water is becoming a scarce resource.
A.I. Is Coming for Lawyers, Again — from nytimes.com by Steve Lohr (behind paywall)
Previous advances in A.I. inspired predictions that the law was the lucrative profession most likely to suffer job losses. It didn’t happen. Is this time different?
Excerpt:
But unless the past isn’t a guide, the impact of the new technology is more likely to be a steadily rising tide than a sudden tidal wave. New A.I. technology will change the practice of law, and some jobs will be eliminated, but it also promises to make lawyers and paralegals more productive, and to create new roles. That is what happened after the introduction of other work-altering technologies like the personal computer and the internet.
One new study, by researchers at Princeton University, the University of Pennsylvania and New York University, concluded that the industry most exposed to the new A.I. was “legal services.” Another research report, by economists at Goldman Sachs, estimated that 44 percent of legal work could be automated. Only the work of office and administrative support jobs, at 46 percent, was higher.
Lawyers are only one occupation in the path of A.I. progress. A study by researchers at OpenAI, the creator of ChatGPT, and the University of Pennsylvania found that about 80 percent of American workers would have at least 10 percent of their tasks affected by the latest A.I. software.
Also relevant/see:
039 | Micro-legal & AI Legal Help — from thebrainyacts.beehiiv.com
AutoGPT might be the next big step in AI.
Here’s why Karpathy recently said “AutoGPT is the next frontier of prompt engineering”
AutoGPT is the equivalent of giving GPT-based models a memory and a body. You can now give a task to an AI agent and have it autonomously come up… pic.twitter.com/mYIJm2IEZy
— Lior? (@AlphaSignalAI) April 11, 2023
AutoGPT, less noise more signal
This is where you should pay attention
Here is why ?
— Linus (???) (@LinusEkenstam) April 11, 2023
AutoGPT is the next BIG thing in AI.
Seems like a new *groundbreaking* update comes out every hour.
Here are a few examples just from the past 24 hours: ?
— Barsee ? (@heyBarsee) April 12, 2023
AutoGPT is the next big thing in AI— from therundown.ai by Rowan Cheung
Excerpt:
AutoGPT has been making waves on the internet recently, trending on both GitHub and Twitter. If you thought ChatGPT was crazy, AutoGPT is about to blow your mind.
AutoGPT creates AI “agents” that operate automatically on their own and complete tasks for you. In case you’ve missed our previous issues covering it, here’s a quick rundown:
-
- It’s open-sourced [code]
- It works by chaining together LLM “thoughts”
- It has internet access, long-term and short-term memory, access to popular websites, and file storage
.
From DSC:
I want to highlight that paper from Stanford, as I’ve seen it cited several times recently:.
- Generative Agents: Interactive Simulacra of Human Behavior
Joon Sung Park, Joseph C. O’Brien, Carrie J. Cai, Meredith Ringel Morris, Percy Liang, Michael S. Bernstein
.
From DSC:
And for a rather fun idea/application of these emerging technologies, see:
- Quick Prompt: Kitchen Design — from linusekenstam.substack.com by Linus Ekenstam
Midjourney Prompt. Create elegant kitchen photos using this starting prompt. Make it your own, experiment, add, remove and tinker to create new ideas.
…which made me wonder how we might use these techs in the development of new learning spaces (or in renovating current learning spaces).
From DSC:
On a much different — but still potential — note, also see:
A.I. could lead to a ‘nuclear-level catastrophe’ according to a third of researchers, a new Stanford report finds — from fortune.com by Tristan Bove
Excerpt:
Many experts in A.I. and computer science say the technology is likely a watershed moment for human society. But 36% don’t mean that as a positive, warning that decisions made by A.I. could lead to “nuclear-level catastrophe,” according to researchers surveyed in an annual report on the technology by Stanford University’s Institute for Human-Centered A.I., published earlier this month.
AutoGPTs are improving at a blazingly fast speed and could soon transform the face of business.
Here's what you need to know:
— Nathan Lands (@NathanLands) April 12, 2023
Credentialed Learning For All — from gettingsmart.com
Vision
Learning happens throughout life and is not isolated to the K-12 or higher education sectors. Yet, often, validations of learning only happen in these specific areas. The system of evaluation based on courses, grades, and credit serves as a poor proxy for communicating skills given the variation in course content, grade inflation, and inclusion of participation and extra credit within course grades.
Credentialed learning provides a way to accurately document human capability for all learners throughout their life. A lifetime credentialed learning ecosystem provides better granularity around learning, better documentation of the learning, and more relevance for both the credential recipient and reviewer. This improves the match between higher education and/or employment with the individual, while also providing a more clear and accurate lifetime learning pathway.
With a fully-credentialed system, individuals can own well-documented evidence of a lifetime of learning and choose what and when to share this data. This technology enables every learner to have more opportunities for finding the best career match without today’s existing barriers around cost, access, and proxies.
Addendum on 4/28/23 — speaking of credentials:
First Rung — from the-job.beehiiv.com by Paul Fain
New research shows stacking credentials pays off for low-income learners.
Stacking credentials pays off for many low-income students, new research finds, but only if learners move up the education ladder. Also, Kansas is hoping a new grant program will attract more companies to participate in microinternships.
How Easy Is It/Will It Be to Use AI to Design a Course? — from wallyboston.com by Wally Boston
Excerpt:
Last week I received a text message from a friend to check out a March 29th Campus Technology article about French AI startup, Nolej. Nolej (pronounced “Knowledge”) has developed an OpenAI-based instructional content generator for educators called NolejAI.
Access to NolejAI is through a browser. Users can upload video, audio, text documents, or a website url. NolejAI will generate an interactive micro-learning package which is a standalone digital lesson including content transcript, summaries, a glossary of terms, flashcards, and quizzes. All the lesson materials generated is based upon the uploaded materials.
From DSC:
I wonder if this will turn out to be the case:
I am sure it’s only a matter of time before NolejAI or another product becomes capable of generating a standard three credit hour college course. Whether that is six months or two years, it’s likely sooner than we think.
Also relevant/see:
From DSC:
Before we get to Scott Belsky’s article, here’s an interesting/related item from Tobi Lutke:
I just clued in how insane text2vid will get soon. As crazy as this sounds, we will be able to generate movies from just minor prompts and the path there is pretty clear.
— tobi lutke (@tobi) March 29, 2023
Our World Shaken, Not Stirred: Synthetic entertainment, hybrid social experiences, syncing ourselves with apps, and more. — from implications.com by Scott Belsky
Things will get weird. And exciting.
Excerpts:
Recent advances in technology will stir shake the pot of culture and our day-to-day experiences. Examples? A new era of synthetic entertainment will emerge, online social dynamics will become “hybrid experiences” where AI personas are equal players, and we will sync ourselves with applications as opposed to using applications.
Unauthorized sequels, spin-offs, some amazing stuff, and a legal dumpster fire: Now lets shift beyond Hollywood to the fast-growing long tail of prosumer-made entertainment. This is where entirely new genres of entertainment will emerge including the unauthorized sequels and spinoffs that I expect we will start seeing.
Also relevant/see:
Digital storytelling with generative AI: notes on the appearance of #AICinema — from bryanalexander.org by Bryan Alexander
Excerpt:
This is how I viewed a fascinating article about the so-called #AICinema movement. Benj Edwards describes this nascent current and interviews one of its practitioners, Julie Wieland. It’s a great example of people creating small stories using tech – in this case, generative AI, specifically the image creator Midjourney.
Bryan links to:
Artists astound with AI-generated film stills from a parallel universe — from arstechnica.com by Benj Edwards
A Q&A with “synthographer” Julie Wieland on the #aicinema movement.
From DSC:
How will text-to-video impact the Learning and Development world? Teaching and learning? Those people communicating within communities of practice? Those creating presentations and/or offering webinars?
Hmmm…should be interesting!