Is investing in AI technology a bet against the odds?

AI investment DXC Blogs

In my last post, I made a big distinction between investments in proven artificial intelligence solutions and investing in growing AI intelligence. While the first situation is relatively safe, the second is a huge risk.

I am not saying that business should stop investing in better AI technology or that progress is hopeless. After all, without effort, there can never be advancements. All the big technological achievements of our civilization (remember people like Nikola Tesla, Tomas Edison and Karl Benz) are based on someone betting against the odds, and succeeding.

What I am saying is that investors need to be aware of the type of bet they are making in AI at this time.

Scaling of intelligence is much harder than scaling the number of users. And this distinction has important implications as businesses attempt to implement AI in their real-world operations.

Consider, for example, a retailer that would like to improve its web services by offering user-tailored purchase recommendations. The company may shop around for an adequate technology and run into a startup with a completely new AI technology that seems better than anything else on the market. The retailer then may be faced with all kinds of decisions. Should it buy the technology? Should it buy the entire startup?

What advice would I give? Well, from what I’ve discussed so far in my blog, the retailer should not be worried about technical aspects of scaling a solution to a large number of users. If the startup has successfully demoed the application, expansion of its use is relatively easy.

However, the retailer should not be tempted to think the startup will develop even better technology in the near future. Someone on the board may say, “Look what they did in the last three years. Imagine what they are capable of doing in three years from now.

Enchanted and eager to beat up the competition, the retailer may decide to acquire the startup and pay a price reflecting the hypothetical value of machine intelligence three years from now. But if the retailer goes this path, it’s clear its board members didn’t read my last blog. 😉

To understand how someone could get swept up in the fervor of AI, just look at this example of coverage in the press:

Most recently, AI has received a great deal of commercial attention. Many large companies are investing significant resources in AI. For instance, about half of the companies in the Fortune 500 are actively pursuing one area of AI.

Some of the companies investing in AI want to create commercial product to sell. More often, companies want to use AI to improve productivity. They believe that AI is a way for them to maintain or capture a strategic advantage within their industry and the marketplace. In other words, companies are investing in AI because they believe that there is substantial commercial value in using machines to emulate portions of human behavior that have not been captured by traditional computing.

The piece ends with this:

We are even more excited that what is currently available is only the shadow of things to come within the next two to four years.

What if I told you this was not written in 2016 or 2015 but in 1990? And what if I told you that this year was the beginning of an AI winter? Needless to say, “the shadow of things to come” never became a reality.

And yet, many startups offer better products than anything that existed before and, by all likelihood, will continue to do so.

What’s important to realize is the hidden ceiling involved. Whatever levels of intelligence your AI has reached, there is no guarantee that it will grow further with just a bit more effort, resources or computational power. In fact, for the technology we use today, there will likely always be a performance ceiling that’s impossible to exceed without some significant change to the technology.

The ceiling is hidden because you don’t know when you will reach it until you do, and it may eventually cause you to fail commercially. Thus, betting on future revenues from continuous improvements in intelligence is  much of a risk.

To be safer with your investments, ask the following questions:

  • Can we roughly assess the growth in resources needed to achieve our planned growth in intelligence?
  • What was historically the exponent for growth in resource demands?
  • What resources do we need, optimistically and conservatively?
  • Do we have a plan B to turn profitable with the existing levels of intelligence?

And do not let your estimates be based on intuition. Let intuition be supported by numbers.

One last thing I would like to suggest is to refrain from claiming that your AI works like a human brain or that you are on the path to creating AGI – and do not believe such claims made by others.

Claims that AI works like human brain are very likely false for the simple fact that we do not yet know how the brain works. Current theories about how the brain function (and there are multiple ones, without consensus) account for only about 10% to 20% of data acquired in physiological research. The remaining 80% to 90% of data is unexplained; nobody knows how to integrate them into a coherent theory.

So to conclude, let me recap the themes I’ve hit on over this series of posts (and please go back and read any you missed along the way):

1 – The AI technology of today is still no match to the intelligence capabilities of a real brain, human or otherwise. What we have is weak AI. And we still have not found a way to begin building strong AI.

2 – As a result, our technology faces the big problem of scaling. Current intelligence technology scales linearly or slower than linearly with the increase in resources. This cannot lead to AGI.

3 – The failure to recognize the problem of scale may have contributed to AI winters in the past, and it could do so again. And, due to the sheer amount of investments in AI at this time, it could bring with it a wider recession.

4 – Therefore, it is crucial that companies investing in AI understand the risks and protect themselves from overpromises and consequent disappointments.

The promise of AI is real and available to us, but only if we chart a wise course to this technology’s future.

RELATED LINKS

Enrolling in Artificial Intelligence Kindergarten

Machine intelligence still requires gray matter

The digital twin

Comments

  1. fishballs says:

    Thank you. A very reasonable and calm assessment of the status quo. I bet that few people are currently interested in hearing realistic assessments of AI, though. We’re in the middle of a boom and the whole thing has taken on a life of its own. People currently want big promises, unrealistic scenarios and phantastic outlooks. So investing in AI is smart now, because everybody wants it and tons of money are being invested by the biggest players.

    Knowing that it is unsustainable will prevent you from holding on to the ride for too long. Just my 2c.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: