AI and the Wizard of Oz

You’ve no doubt heard or seen the scary news that AI will take our jobs. You may have even felt some anxiety at the thought of losing your job to a machine.

But, far from AI taking human jobs, it seems that humans are taking AI jobs.

In recent years, some ‘AI businesses’ have adopted a Wizard of Oz approach to their AI deployment. That is, they’re hoping you’ll pay no attention to the human behind the curtain.

But what does this deceptive AI practice mean for real artificial intelligence?

The Wizard of Oz in AI

The Wizard of Oz approach to AI is where a business uses human labour masquerading as artificial intelligence. It’s a form of fake AI (or ‘AI snake-oil’) caused by deliberate deception.

Instead of robots pretending to be human, humans are hiding behind the curtain, pretending to be robots.

For instance, this year a survey found that approximately 40% of European AI start-ups don’t use AI. Humans have pretended to be chatbots. They’ve even acted in place of machines to transcribe voicemails and other sound files into text.

Why is it happening?

Some fake AI is born of genuine misunderstanding. That is, confusing automation software with AI, or suffering a misclassification from a third party. But Wizard of Oz AI is different because it’s fake on purpose.

One reason behind the Wizard of Oz approach to AI is a desire to cash in on the hype of the technology. Investors are particularly interested in artificial intelligence. So much so, that an ‘AI’ classification can attract a funding increase of as much as 15 – 50%.

A (slightly) more legitimate reason for hiding humans behind the AI curtain is to test waters. So, a company that has every intention of building and using AI technology is prototyping the tech.

Artificial intelligence is expensive; it takes masses of data to train it into a useful tool. By testing the waters with humans, AI start-ups and businesses ensure there’s a demand for that kind of AI functionality. (Before they invest time, effort and money into it.)

Ethical implications and AI acceptance

Despite the obvious ethical concerns surrounding Wizard of Oz AI, there are potential positives to the practice.

For a start, introducing artificial intelligence before it exists could highlight ethical concerns early-on. Think consumers taking to Twitter to voice their concerns, or customers leaving feedback that they distrust the new ‘AI’ agent helping them. This would allow brands to address AI concerns before they’re an issue. In turn, this could help encourage AI acceptance. So, it’s easier to adopt AI once the technology reaches the target ability.

That said, the practice brings its own ethical concerns. Alongside the flagrant deception, there’s a concern about privacy. Think AI tools used for mental health or medical support. Or humans listening to your conversations — even those you thought no one would hear. This hurt trust could hinder AI acceptance and future implementation, rather than help it.

Demand or disillusion?

Long term, this practice could shape future uses of AI. It helps AI researchers determine where there’s the most need for the technology. So, they then direct development to those areas.

But the practice risks severe disillusionment. As people discover that their AI-powered tools are fake, they will inevitably grow distrustful of the technology. As such, when they encounter a real AI tool, they won’t trust it.

Plus, artificial intelligence still has a long way to go before it reaches human levels. Humans pretending that they’re bots could over-elevate the tech’s ability, lessening the accomplishments of ‘real’ AI.

All this distrust and disillusionment could then feed into another AI winter. That is, a period of low advancement and hushed development. ‘Artificial intelligence’ would become a dirty buzzword.

The AI Wizard of Oz

There’s no doubt that hiding humans behind an AI curtain is an unethical practice. Actively deceiving people is always going to come with an ethical question mark attached. No matter the reasoning behind it.

As technology grows into the roles held by the humans behind the curtain, there’ll be little need for Wizard of Oz AI.

For now, it’s disheartening when the curtain comes away to reveal it’s all a hoax. But only time will tell how damaging a practice it is for AI advancement.

Note: we originally published this article here: