Week in Edtech, 12/12: Edtech Meets Generative AI
What might ChatGPT, Midjourney and more enable in edtech?
DEEP DIVE: WHAT MIGHT GENERATIVE AI MEAN FOR EDTECH?
Over the last couple of weeks, you couldn’t swing a cat without hitting another hot take about Generative AI- specifically, the newest AI algorithms that create complex and convincing text and/or images from the simplest of human prompts.
In the last few months, we’ve all been introduced to a world in which artificial intelligences can, in response to simple inputs, create original and unique images like this:
In fact, in recent months we’ve learned that AI can effectively create original:
Images: Image creators like Midjourney and OpenAI’s DALL-E 2 can create a bewildering array of images in virtually any style, almost instantaneously, including logos, cartoons, artistic styles or photorealistic images.
Text: OpenAI’s ChatGPT 3.5 in particular, has seen a meteoric rise in usage, reaching over 1mm users in less than a week after launch with its ability to generate textual responses.
Audio (Music and Voices): Other uses are close behind, like Google’s AudioLM, which has the ability to hear a snippet of a song and continue playing it or to rapidly learn voices and then create original, modulated voices with emotion, pitch and speed.
3D Models: Nvidia just announced Magic3D, which can generate 3d models of objects from simple text descriptions.
For a more thorough overview of the tools out there, here’s a useful list of Generative AI tools of all kinds from researcher Steven Van Vaerenbergh.
According to modern convention, this would be the moment in which I say “surprise, the last few paragraphs were written by ChatGPT”! But alas, no- nobody here but us humans.
But What Does it Mean for Edtech?
I’m pretty late to the game here; there have already been a number of great pieces written about what this moment might mean for Edtech, from Edtech luminaries like:
GSV’s Michael Moe in his EIEIO newsletter
Transcend Networks’s Alberto Arenaza in his Substack
Reach Capital’s Tony Wan
Hanna Celina, whose first hand account of creating educational content with ChatGPT was cited in this newsletter last week
In the Atlantic, Educator Daniel Herman claims it’s the “End of High School English”; professor Ian Bogost says it’s “dumber than you think”, prone to recreating five-paragraph essays… there’s even a hot take on AI in Edtech from Atlantic CEO Nicholas Thompson, who predicts that AI will upend homework and essay writing as we know it.
All I can add to this discussion is that, to me, this feels like one of the few real ‘wow’ moments in technology- as transformative as, say, the original iPhone with its selfie camera, or the first social networks, or the sudden availability of streaming video.
Just as we had our Web 2.0 revolution with Flickr, Youtube and the classic “Myspace vs. Friendster” debates in the early aughts, this is our AI 2.0 moment- the point in time in which artificial intelligence suddenly goes from an abstruse tool used by geeks to one accessible to absolutely everyone, and that will be used in ways we can’t even imagine.
Moreover, I would venture that the slate of ‘problems’ that folks are calling out: that AI:
is shallow, regurgitates nonsense and doesn’t understand what it’s really saying
it’s often flat wrong or at best, “bullshitting”
it basically invites cheating at all levels by answering any question or even writing cohesive essays and papers for students
it allows massive uses disinformation and deep fakes, exposing the need for digital literacy
it can even undress people without their consent
… are be missing the forest for the trees. For one thing, all of the claims above- yes, every one- are also made about the internet culture itself- and yet… it has fundamentally transformed how humanity solves problems and spends its time.
Secondly, the technology is still quite nascent, and let’s not up and forget the actual power of AI- massive data-based reinforcement learning, which mines each and every data point to continually improve its models, identify new features, and sharpen outputs.
These incredible technologies already allow dyslexic learners to use chatbots to communicate, or enable eleven year olds to hold GPT-3 hackathons (via Mindjoy). They already allow us to translate 200 languages in real time, or compose original music in a wide variety of styles.
For educators, it means that AI is already writing educational raps about quantum physics in the style of Snoop Dogg:
… and it will soon be able to actually perform the raps, in Snoop Dogg’s voice, with its own composed music.
It means that AI is already designing entire movie synopses and scripts, complete with AI-generated actors…
…but soon, it will be able to generate the entire movies from scratch.
Let’s be clear here: any individual can now create professional quality illustrations, art images, photos, text, poems, worksheets, music, code, speech, 3d objects, and even games and video.
Not only can educators instantly create materials for any learner, in any subject or style, for any topic or interest… but learners will be able to create their own materials.
Whew- I’m getting a bit overheated here- will have to save the rest for part 2, in which
BIG FIVE HEADLINES
1. EDTECH ACQUISITIONS REVEAL EMERGING STRATEGIES AND STRUGGLES
A number of high-profile acquisitions reveal strategic initiatives for Edtech companies, and show that the runway for smaller Edtechs is starting to run out.
Top Hat acquires chemisty platform Activ Learning for extended reach
European Edtech Unicorn GoStudent acquires In-Person Tutoring Centers to hedge against the return to in-person learning
Simplilearn acquires Fullstack Academy to branch into bootcamps
2. INVESTMENT MONEY HEADING TOWARD UPSKILLING
3. EDTECH STOCK WOES EXPLAINED
Phil Hill’s two part series first analyzes the losses in public Edtech value since the Coursera IPO and then posits several drivers of this loss
We have two phenomenal interviews this week, both on the hugely important topic of Edtech and Special Education:
Kate Eberle Walker, CEO of Presence (and former CEO of Princeton Review and Tutor.com), about the stellar rise of teletherapy for students with diverse needs.
Special Education experts Yair Schapira, Judy Rich, Luann Purcell and Angelica Morgan of Amplio Learning and CAST
Thanks for reading Edtech Insiders! Subscribe for free to receive weekly updates and support the podcast.