TEDTalk Share - Recent AI Talks

Daily Resource Share / May 3rd, 2023

There is just so much out there on artificial intelligence right now. Just spending a few minutes exploring social media and what is being said by experts, armchair experts, and laypeople alike can make it hard to discern the best responses to AI’s boom over the last few months. Since the release of ChatGPT-4 in November and the extreme proliferation of AI-based models for a variety of tasks, there have been too many resources to turn to as an educator that wants to learn the possibilities of AI and how it can influence education. The AI rabbit hole is deep, wide, and full of inaccurate information.

So, like with many topics, I turn to TED. In the last week they have been releasing the Talks from their TED event this past month and no surprise, a few of the talks have been directly on AI. First off, we have Professor Yejin Choi, whose work is in machine learning and natural language processing. She makes an incredible case for why AI is not as intelligent as we think it is by exploring the work of Sun Tzu. By knowing our enemy, choosing our battles, and innovating our weapons, we can see that AI is not as intelligent as it seems and that we need to rethink how we are developing AI. It is a fascinating argument that helps to build a better understanding of how AI is currently trained, why there is a call to slow down and think through research, and, at its core, what AI actually is and how it functions in our world today. There is a great point where Professor Choi points out that common sense is a key component missing from AI, that we might be able to train it using 1 Trillion words, but if we can’t get it to use common sense to answer questions then AI is not functioning the way we want it to be functioning. She also makes a great series of points about the data that is training AI and how if we choose “garbage in, garbage out” AI will not be appropriately trained.

On the other hand, Greg Brockman, OpenAI’s cofounder, offers a vision of AI that will in the coming months: a web-connected, super assistant that will ask permission before sending, ordering, or sharing. The connections between applications are very interesting and could help create a workflow that many of us could be helped by. I appreciated his use of recipes to shopping lists to ordering food to sharing online workflow. Although many of us will still choose to shop the traditional way, a person who is homebound could benefit deeply from a digital assistant that helps them go from recipe to delivered goods. Or for someone who is learning a practice and is unsure how to start, AI can help to find and develop a list of goods to meet those needs. I am curious as I was thinking about this talk of the last week or so, how this could affect DIY projects at home or simple instruction manuals. It is a vision of AI that is helpful and integrated into our lives in beneficial ways. It made me think back to the first Apple presentation of the iPod Video. A co-worker at the time completely dismissed videos being watched on a screen that size. Now that is the norm for video watching and Instagram Reels and Youtube Shorts are changing how we view videos again.

If Professor Choi’s talk was a much-needed reality check, Greg Brockman’s talk was a picture of what can be when Professor Choi’s points, among other researchers’ points, are examined, integrated into the discussion, and deployed into AI. Check out both videos to explore the ideas for yourself.

Previous
Previous

UEN Homeroom on AI in Education

Next
Next

The AI Frontier Wakelet Collection