Back in 1950, Alan Turing repositioned the question “Can machines think?” as an engineering challenge. He called the challenge “the imitation game,” popularly known as the Turing test.
To pass the Turing test, machines need to imitate human thinking and prove it to a human judge through open-domain conversation. This set off decades of research and innovation in artificial intelligence (AI) and machine learning (ML).
Turing’s impact is undeniable, but how relevant is imitation today?
In recent years, the business community has started to weigh in more heavily on AI, and in many ways, that’s shifting the conversation from what’s possible to what actually generates value.
Would it be better for businesses to forget about the question of imitation? Could we even use the differences between human and artificial intelligence to our advantage?
I think it’s time businesses really double down on domain-specific AI. Here’s why.
Why the imitation game is bad for business
At the time, the imitation game seemed like a practical approach to a philosophical question. Turing predicted in his 1950 paper that by the year 2000, advancements in computing power would enable machines to pass the test, and people would no longer consider “machine thinking” a contradiction.
The adaption of analytics is increasing rapidly. Computing power has grown and the fields of AI and ML are booming.
However, there is still a lot of suspicion around machine intelligence, and the fact is – today’s machines aren’t all that good at open domain thinking.
For example, in a conversation about Leeds, the football team, a machine may interpret the question, “How is the season going?” as a request for the weather in Leeds, England. That confusion was a dead giveaway in 2016, when a human judge was assessing the machine Mitsuku.
Modern machines can’t drive as well as humans either. The Washington Post recently published footage of full self-driving car testers, and it’s not pretty. Experts explain why it’s so difficult to get right, “where patching one issue might introduce new complications, or where the nearly infinite array of possible real-life scenarios is simply too much for Tesla’s algorithms to master.”
Businesses are right to be wary. There’s a lot out there that AI can’t do yet. Just like how bad full self-driving cars could be extremely dangerous on the road, bad algorithms deployed within business workflows could be extremely damaging to business.
If AI can’t imitate humans, what is it good for?
AI that can fully imitate humans is considered general AI. General AI would exhibit cross-domain thinking and teach itself new strategies. It would definitively pass the Turing test.
Scientists and engineers should go for it. Keep working on self-learning algorithms and pushing the limits on technology. However, where business is concerned, we can’t put all our bets on the widespread availability of general AI in the next 3-5 years.
Luckily, we don’t have to. There’s a huge body of work on AI and proven business use cases that we can start putting into practice immediately.
Today, AI is more popular than ever in the business world. Corporate investment in AI increased by 5x from 2015 to 2020 (Stanford), 79% of companies are exploring or piloting AI projects (Gartner), and 86% will consider AI a “mainstream technology” (PwC).
If AI isn’t good at open domain intelligence, then what is it good at, and what’s the market reflecting?
Here’s your answer – it’s domain-specific AI.
The advantages of domain-specific AI
According to Lee Kaifu, a computer scientist and thought leader in AI, “Today’s A.I., or what we call A.I., is actually very narrow, domain-specific, but incredibly capable and superhuman within very limited tasks.”
Domain-specific AI can automate tasks that humans do, but the focus isn’t to imitate humans. It won’t be able to switch from one task to another and then dive into a conversation with you about its plans for the weekend.
However, in certain well-defined tasks, domain-specific AI wildly outperforms humans. Imagine your colleague is now summarizing thousands of articles a day. That would not be an imitation of normal human behavior.
Moreover, true domain-specific AI, that incorporates domain knowledge, is smart at that task – maybe even smarter than you. It’s infused with expert-level domain knowledge, let’s say in pharmaceuticals, so it can summarize heavy research papers on drug trials much faster and potentially better than your average college graduate who majored in biology. The advantages of domain-specific AI are not only speed and scale but also the ability to distribute highly specialized knowledge so it can be put to use.
Evalueserve has a model risk management product for banks that can accurately interpret model test results and generate proper documentation – work that’s normally done by highly educated quants, often PhDs.
The benefits of domain-specific AI, in this case, include not only scaling the capacity of the model risk management team but upskilling junior team members, giving them the ability to perform more complex tasks.
Forget about imitation and embrace domain-specific AI
While the Turing test may have revealed that AI as a field is still quite immature, domain-specific AI is definitely a viable and mature technology that businesses can and should consider implementing now.
Domain-specific AI fits right into our existing business processes, especially where organizations are already using data to drive decisions and automate tasks. It helps solve a lot of existing pain points such as talent gaps and the need to process and analyze data at scale.
So as business users, let’s forget about imitation for now and embrace the benefits of domain-specific AI.
After all, do we even want to unleash AI that fully imitates humans right now? I’m not sure society is ready, let alone our existing business processes and technology infrastructure.
To talk to an expert about domain-specific AI, you can contact us here.
