Artificial Intelligence is Not the Silver Bullet for K–12

artificial

Artificial intelligence is receiving a great deal of press recently — for good reason:

  • "How the Computer Beat the Go Master": While chess fell to a computer in 1997, when Deep Blue beat the World Champion, Gary Kasparov, the board game "Go" has proven more elusive. Why? Go has many more "states" than chess and thus the game playing strategies needed to be improved, and maybe even more importantly, the computers simply needed to be much faster. The Google company, DeepMind, took care of the former issue, while good old Moore’s Law took care of the latter issue.
  • "Inside the 20-Year Quest to Build Computers That Play Poker": And poker is more challenging yet! In chess and Go there is "complete information" — each player knows all the information about the board and the potential moves. In poker, however, each player doesn’t know what cards the other players have in their hands and they don’t know what cards will be dealt by the dealer. Poker is a game of "incomplete information." But yet, now poker has fallen to the computer!

And we hear that major companies are making major investments in AI:

  • "Nine tech giants investing in artificial intelligence: Microsoft, Google, Uber and more are investing in AI"
  • "Inside Look: The World's Largest Tech Companies Are Making Massive AI Investments"
  • "IBM to invest $1 billion to create new business unit for Watson"
  • "Ford is putting $1 billion into an AI startup, Detroit's biggest investment yet in self-driving car tech"
  • "AI Partnership launched by Amazon, Facebook, Google, IBM, and Microsoft"

Still further, there is even an organization — a nonprofit — that is concerned with the potential negative impact of AI on society.

Clearly, AI is no longer just an ivory tower discipline; it is fast becoming a major economic and social force.

No surprise, then, that the following question has been raised:

  • Statement 1: "Is artificial intelligence the next big wave in ed-tech offerings?"

And, given the above context, it is no surprise that the response is:

  • Statement 2: "The question of artificial intelligence in education is not really one of 'if' but of 'when.' "

Indeed, artificial intelligence already powers the "personalized learning" systems produced by several companies (e.g., Dreambox, Knewton).  And, studies are showing increased student achievement when students use those personalized learning systems.

Yeah, yeah, yeah. But, we still stand by the title of our blog: "Artificial Intelligence Is Not The Silver Bullet for K–12." Strap in! We will be moving very quickly in the next few paragraphs; we apologize for the vertigo that might well be induced.

Step 1: An important distinction needs to be made: Artificial intelligence is not "machine learning." Broadly speaking, there are two types of artificial intelligence: "Symbolic AI" and "Non-symbolic AI."

  • Symbolic AI: Symbolic AI is characterized by systems that explicitly reason using logical rules, e.g., All birds are mammals, a robin is a bird, therefore, by the logical rule, "modus ponens," we can conclude that a robin is a mammal. Of course, in real symbolic AI systems, the reasoning will get much more sophisticated, nuanced, complicated.  Symbolic AI underlies some of the "intelligent tutoring systems" that have been built.
  • Non-symbolic AI: One form of non-symbolic AI is called "machine learning." This type of AI uses a range of algorithms that are based, by and large, on probability and statistics. Machine learning algorithms look for statistical correlations — machine learning is really machine training. Non-symbolic AI underlies the "personalized learning" systems used in classrooms.

We refer the interested reader to several articles where this distinction is explored in much greater detail.

Step 2: The dramatic progress in and impact by AI being referred to in the first paragraph, above, is in non-symbolic AI, in machine learning. Symbolic AI — the type of AI that Cathie and Elliot (C&E) did during their graduate school days in the dinosaur age — is still an academic discipline, by and large.  Thus, statements 1 and 2 really should read:

  • Statement 1: "Is machine learning the next big wave in ed-tech offerings?"
  • Statement 2: "The question of machine learning in education is not really one of 'if' but of 'when.' "

And thus the title of this blog, then, really should read: "Machine Learning is Not The Silver Bullet for K–12."

As we noted above, sports fans, we are covering a lot of ground, very quickly! Each one of the above steps is worth 1000 times the space we have given it here. But, as this is a blog, not a treatise, so be it.  

Step 3: We stand by the title of the revised title of our blog: "Machine Learning" is Not The Silver Bullet for K–12.

Why?

  • Machine learning powers personalized learning.
  • Personalized learning is not the silver bullet for K–12. (See expanded argument, below.)
  • Thus, machine learning is not the silver bullet for K–12.

Expanded argument:

Now, IOHO, personalized learning is not the silver bullet for K–12. As we commented in an earlier blog, personalized learning does have a place in the educator’s tool box. When the goal is to teach a fixed set of procedures, e.g., in early math, then personalized learning, a form of direct instruction, can be effective.

The key to one, but nonetheless representative, form of personalized learning system uses machine learning to carry out:

  • "Continuous formative assessment in and between lessons, analyzes over 48,000 data points per student, per hour to provide the right next lesson at the right time."

Yes, machine learning techniques can deliver just the right problem to a student at just the right time which may be an appropriate instructional strategy to teach arithmetical procedures.  

But wait, who should be doing the continuous assessment? The computer? John Dewey, one of the world’s pre-eminent educational philosophers, would argue that it is the learner who should be doing the continuous assessment.

  • "True reflective thinking requires the continuous re-evaluation of ambiguous or unsettled situations as beliefs, assumptions, and hypotheses shift." (A paraphrase of an aspect of Dewey’s educational philosophy.)

What does one reflect on, then?

  • "Failure is instructive. The person who really thinks learns quite as much from his failures as from his successes." –John Dewey

Reflecting thinking — "continuous formative assessment" — is a key skill that is learned in the context of classrooms that employ science curricula based on the Next Generation Science Standards (NGSS). One of the central goals in NGSS is having learners "figure things out," e.g., why do earthquakes stop? By definition, "figuring out" means trying different understandings, reflecting and trying again. Direct instruction — embodied in personalized learning systems — surely is not an appropriate pedagogical technique to help learners develop figuring out skills!

NGSS deals with science. "Big History" is a new way to teach/learn about history. Key to Big History is also the notion of "figuring things out." Language arts? Writing well does require an understanding of grammar. Does that mean drilling children on the grammar rules first via personalized learning systems powered by machine learning algorithms, is a good pedagogical practice? Excellent question! As the personalized learning systems branch out into teaching grammar, we look forward to reviewing the research.

In sum, then, we would argue that only a small percentage — a very small percentage — of what is to be learned (e.g., early math procedures, maybe grammar rules) in K–12 can be taught using personalized learning systems powered by machine learning. Now, silver bullets are meant to address big problems, big percentages. Thus … sound the trumpets …" artificial intelligence in the form of machine learning is not the silver bullet for K–12."

Heck of a ride, eh?! 

Featured

  • glowing digital human brain composed of abstract lines and nodes, connected to STEM icons, including a DNA strand, a cogwheel, a circuit board, and mathematical formulas

    OpenAI Launches 'Reasoning' AI Model Optimized for STEM

    OpenAI has launched o1, a new family of AI models that are optimized for "reasoning-heavy" tasks like math, coding and science.

  • landscape photo with an AI rubber stamp on top

    California AI Watermarking Bill Supported by OpenAI

    OpenAI, creator of ChatGPT, is backing a California bill that would require tech companies to label AI-generated content in the form of a digital "watermark." The proposed legislation, known as the "California Digital Content Provenance Standards" (AB 3211), aims to ensure transparency in digital media by identifying content created through artificial intelligence. This requirement would apply to a broad range of AI-generated material, from harmless memes to deepfakes that could be used to spread misinformation about political candidates.

  • clock with gears and digital circuits inside

    Report Estimates Cost of AI at Nearly $300K Per Minute

    A report from cloud-based data/BI specialist Domo provides a staggering estimate of the minute-by-minute impact of today's generative AI boom.

  • glowing lines connecting colorful nodes on a deep blue and black gradient background

    Juniper Intros AI-Native Networking and Security Management Platform

    Juniper Networks has launched a new solution that integrates security and networking management under a unified cloud and artificial intelligence engine.