2018 AI predictions: Summary of top AI experts’ predictions

2018 AI predictions: Summary of top AI experts’ predictions

Since the beginning of the year, PwC, CEO of pymetrics, Gil Press  published predictions on the direction AI will take. We read them all and couldn’t resist adding our predictions and categorizing the predictions:

Mega-trends that will shape AI in 2018

The news cycle is full of AI, research centers being opened, re-organizations, new research findings and tabloids peddling that robots will kill us all tomorrow. Reading a different type of news everyday, it is easy to lose track of what is really happening. What are the major trends?

My initial aim was to try to keep predictions as specific and measurable as possible so when we look back we can understand if the trends were really there. However, this is hard to do without a serious research effort so I decided to sacrifice some of the measurability of my predictions.

I am taking a geographic approach in the predictions because AI is shaping different countries differently. While loss of jobs is the prevailing fear in North America and Europe, rise of an unbeatable surveillance state is a phenomenon that is already affecting Chinese.

Trends in Europe and North America

Increasing hype around AI

Though it may feel like we hit peak AI, I think we are still not there. I expect to continue to see puppets masquerading as sentient beings, buzzword heavy initiatives like distributed computing with blockchain proof of work tokens to revolutionize AI model development.

Ok this is not measurable but I expect to see more buzzwords crammed into startups.

Increasing automation

Sadly, most white collar work does not really require a university degree. Strong common sense, basic algebra and planning skills are enough to complete most tasks. Intelligent automation systems are already relieving people of work that is not complex enough to require a human’s attention. I expect more companies embracing automation solutions and increased valuations of automation companies such as robotics process automation or robotics companies.

Increasing ROI pressure

While big tech is spending hundreds of millions in Artificial General Intelligence (AGI) and AI research, AGI is at least decades away. And AI projects that fail to deliver results in areas like content moderation are resulting in embarrassments for major brands like Youtube.

I expect heavier focus on ROI especially from companies other than tech giants and to see more news like MD Anderson Cancer Center spending millions on an AI project and canceling the project without achieving intended results. AI companies will need to highlight their ROI more clearly which is a challenge especially in projects for support functions like HR.

Increased fear of AI

Post-2018 election, tech companies are no longer seen as our saviours. They are more like bankers of the 80s, doing anything to get more power and money. And people are right to be afraid. It is now almost universally accepted that millions of people will need to be retrained as AI systems take over their current jobs.

Biases in AI systems is another source of fear. The most important decisions in our lives like getting hired or approval for our credit application are getting partially or completely automated. We will naturally question our changing relationship with machines. Once mere tools, now they emotionlessly shape our lives. To combat this, I expect increased search on and further breakthroughs in understanding the black box of neural networks. We need to understand how machines think if we start allowing them to make decisions that impact lives.

Increased calls for accountability by AI companies

Most of the time, emerging tech gets a free pass. Facebook was used to mine detailed profiles of millions, twitter served as a haven for hate speech… AI companies have also been getting a relatively free chance; relying on opaque and brittle systems making important decisions. However, there will be backlash if incidents like Uber’s self-driving vehicle’s accident continue happening.

Russia and China

Militarization of AI

AI is already eating cybersecurity. Most cybersecurity companies founded in the last 5 years heavily use AI. A prominent example is Darktrace which raised almost $200M since 2013.

Real battlefields are also getting automated. While drones have been flying autonomously for a long time, I expect land based deadly drones as well. Though prominent scientists and entrepreneurs are working to stop this deadly trend, sadly human history shows that no improvement in war can be shelved. While we have managed to ban some moderately effective weapons like blinding lasers or chemical weapons from warfare, no weapon that gave significant benefits to its users has been banned. Even if it was banned, in the first major war, we would see the desperate side resorting to it. So I would expect to see especially Russian military rolling out new systems with almost fully autonomous capabilities.

Sadly, war is not the only area where AI can be militarized. China is already conducting massive scale surveillance and its surveillance capabilities can only grow, increasing the stability of the regime.

Rise of China as the world’s AI superpower

There are 2 critical factors to generating increased machine intelligence. Data and scientists.  Scientists design the learning structures which use computing power to learn from data. China has an advantage in both categories.

China has a data advantage. Chinese population is larger and more tech-savy than any Western audience. As a result, China has the world’s largest e-commerce market, mobile commerce market and Wechat, the application that offers the most comprehensive set of services from ordering taxis to paying for your meal. To top this, China has weak data protection laws which allows companies to be more flexible in using the data.

China is already leading or close to leading the world in AI research. China leads the world in number of science and engineering publications and the ratio of Chinese publications that are among top 1% in terms of citations has been increasing. In addition, according to qualitative observations at least, Chinese scientists and engineers are working far more than their counterparts in US or EU. Additionally, Chinese AI startups’ shares are flying through the roof, helping them invest more.

So why did we not hear about these Chinese AI companies? China is more culturally more inwardly focused so Chinese companies tend to start international expansion later than US or EU companies however I believe Chinese AI companies are ripe for international expansion. I expect AI focused tech giants like Alibaba and Baidu to make large acquisitons in EU and SEA while Chinese companies open up their products to global users.

Solving intelligence is one of mankinds biggest challenges, bigger than Manhattan Project and Apollo Program combined so it is naturally hard to predict its future. However, given the stakes it is only natural that billions will continue to be invested in this endeavor.


Leave a Reply

Your email address will not be published. Required fields are marked *