My motivation for working in Education - Part 2: The business perspective
Hi All!
It was a lovely weekend of cooking up new features for Trailblaze on the a16z & Mistral AI hackathon, and great to make new friends in the AI community :)
Trailblaze Platform updates:
Business wins:
Ask:
The business case for AI in Education
There has been a lot of talk recently about whether Education technology is investable from a Venture Capitalist (VC) perspective, and whether it is even a good business. The fall of Byju’s and the end of COVID funding for US schools are scaring off investors, whilst schools seem to be returning to pen and paper after the blip of online schooling during Covid.
Following on from last week’s post, I’d like to argue the opposite, and to dive into why now is in fact the best time to either invest in Edtech or start an Edtech company in the next 20 years. And yes, it’s because of AI.
AI creates a fundamental change in education delivery
Education is uniquely positioned to benefit from AI. To understand this, it’s only necessary to look at one fundamental shift which has already occurred: We can now go from 1 teacher for 30 students to 1 AI teacher for 1 student, and this will have reverberations and ramifications across the industry for the next 2 decades.
Evidence backs up this claim of AI tutors’ potential effectiveness.
The challenge lies in embedding AI tutors into everyday school and home learning processes, demonstrating clear improvements in performance whilst maintaining the socio-emotional and physical education that a school provides, and the personal relationship and motivation to learn that students experience with good teachers.
领英推荐
So, how do we do this?
Slowly, Slowly
To get to a fundamental change in the way education is delivered, we will have to start small and embed AI into existing processes in traditional classrooms and home environments. This way we will maintain the socio-emotional and physical aspects of education by default.
Solutions such as Magicschool are showing how to do this for teachers. Tasks such as lesson planning, slides generation, student feedback and differentiation are tasks that teachers need to do today, and can be done faster with AI.
For students, solutions such as Khan Academy’s Khanmigo, or even DuoLingo’s ‘video call’ feature, are embedding AI features into existing tools that kids use at school, in order to provide a more personalised experience. At home, tools such as Studdy let you take a picture of your current exercise and ask AI for help in solving it.
Evidence shows that this adoption and engagement is happening fast. More than half of secondary students already admit to using ChatGPT-like tools to do some or all of their homework, and 60% of US teachers are using AI tools as part of their workflow.*
And then all at once.
Once we have that baseline level of engagement with and trust in AI, the real opportunity for Education technology in the next 20 years then lies at the moment when education goes from AI as a supporting function to AI as a core function.
This is already being imagined and tested. Existing schools, private companies (e.g. Synthesis school & Eureka Labs) and whole governments (e.g. South Korea) are creating new curricula with AI at the core of the teaching and learning experience.
The gradual adoption of AI in schools that we will have seen from smaller ‘incremental’ tools prior to these new AI-first teaching methods will be fundamental in the AI teaching methods’ success, as they will enable the smoother bridging between, and adoption of, the methods by existing traditional schools.
This will be messy with many failures, but there will be successes, and the successes will be huge in terms of impact and ROI. This is firstly because personalised tutoring is known to provide an improvement of 2 standard deviations in student academic performance, and 50% of parents across the world are already willing to pay for home tuition for a fraction of that improvement. And secondly because the best AI tutors will need the best data about student performance, and data aggregation requires standards, trust and security (especially with children’s data) and is a winner takes all market.
Conclusion
In conclusion, the difference in the quality of academic education available from AI tutors is simply too large that this is a case of when, not if, it is adopted, and the school system will adapt around it. The fundamental shift from 30:1 to 1:1 will have far greater reverberations across the school system than we have seen since the industrial revolution, and this makes now the best time to invest in or be a part of the small companies with the flexibility to see and seize those opportunities early.
All the best,
Hugo
*See a list of sources for more detailed statistics across teachers & students in primary, secondary & tertiary education compiled here.