Reverse Turing Test | South by South West

Thus far, AI has been trying to prove its ability. Soon humans will need to prove their humanity and their abilities

First off, thanks to many of you for your emails in response to my last post on AI and human consciousness. Based on the interest, I plan to continue that series (coming soon).

In 1950, the foundations for AI were laid when mathematician Alan Turing published a paper posing a simple question, “Can machines think?” Turing imagined a scenario in which a computer might chat with humans and trick them into believing that it, too, was human. This hypothesized imitation game, known as the Turing Test, established an ambitious milestone by which to measure the intelligence of machines thereafter. AI researchers often apply versions of this test to other settings from AI enabled writing to art — can a random user separate AI-created work from human work.

As AI is becoming more and more advanced, I think a reverse Turing test is coming our way. This test will involve two related things: (i) Prove you are a human and not AI (ii) Prove you can do a task better than AI.

Proving You’re Human, Not AI

As AI bots increasingly interact with the web under the guise of humans, system administrators will need reliable ways to verify that we are, in fact, human before granting access. Captchas and ReCaptchas are early examples of this, but as AI grows more sophisticated, these tests will need to become more challenging to ensure that only humans can enter certain digital spaces. For instance, if you’re interviewing a job candidate on Zoom, how can you be sure it’s not a digital twin with instant access to the world’s knowledge? We’ll need new methods to confirm that the person on the other end is genuinely human. At least until we have AGI, it will be a cat-and-mouse game —creating increasingly difficult “reverse Turing” tests that humans can pass but remain beyond the reach of AI (and, let’s be honest, even Captchas are getting tough—I often struggle with the “I am not a robot” test myself).

These tests won’t be limited to digital interactions. How will we distinguish between AI-made and human-made products? And does that distinction matter? Studies suggest that people tend to prefer human-made products, at least in certain contexts. Art is a prime example. If handmade art commands a premium, there could be an incentive to falsely label AI-generated art as “human-made.” This mirrors the current value placed on handcrafted products over those made by machines. Therefore, the ability to credibly signal whether something is human-made or AI-made will become more important.

Proving You Can Perform a Task Better Than AI

In many ways, the scenarios above are also about proving that humans can outperform AI in specific tasks. Captchas, for instance, are tasks that humans can currently do better than AI, serving as proof that we’re not robots. But as AI becomes more embedded in the workplace, companies will need to rethink how work is designed. They’ll have to decide which tasks to automate (hand over to AI) and which ones to augment (pair humans with AI tools). Ultimately, this will come down to understanding what humans excel at. For example, my startup collaborates with Hollywood studios and agencies, helping them manage incoming scripts and books. While AI excels at summarizing scripts and handling a massive influx of content thereby ensuring that everything gets read, humans are better at interpreting the emotional impact of a script—how it makes us feel.

I Need Your Help (5 Minutes to Vote for Our SXSW Panel)

We’ve proposed a panel for South by Southwest titled “Redefining Storytelling with AI in Hollywood and Beyond.” This discussion will explore how the fusion of AI and human creativity is opening up new artistic frontiers. If this sounds interesting, I’d be grateful if you could take a moment to vote for our panel by August 18. As I’ve mentioned before, while AI presents challenges for creators, it can also enhance creativity when used correctly. Our panel aims to delve into how we can safely integrate AI into creative processes.

Vote here: SXSW Panel Picker.