Thread by @CSISIndonesia

Good morning! We are now move on to our first plenary session, "The New Technology Frontier in Big Picture". This session moderator is Don Campbell, PECC Chair, Canada. His opening remarks emphasize on the future relationship between human and artificial intelligence.

Our keynote speaker for this session is Luke Hutchison, he is GLG Council member and founding member of Kurzweil Google AI Lab. The title of his speech is "AI and its limitation", which gives the pragmatic view of machine learning.

The secret of happiness is to set low expectation. So, if we start with low expectation on what AI will be capable of, then hopefully we will be pleasantly surprised.

Many people expect that AI will threaten humanity. Yet, he works with 300 of the best minds in AI in google, and nobody has a clue on what intelligence really is or how to build it.

He gave us reality check on the hype of AI versus its reality. If you see AI as reality, these are the 5 things that we can build on AI:

1 Supervised learning: Statistics and regression
2 Unsupervised learning: Clustering the data and take an inherent structure from the data set

3 Reinforcement learning: Time spent between exploration of the world and exploitation of the robot's current ability
4 Symbolic reasoning: Graph search system
5 Other statistical techniques

All of these are necessary but insufficient to build the AI.

Problems that acquire actual intelligence to solve need our whole list of things that we don't know how to build right now. Top problems in the list are:
1. Pragmatic problem solving
2. Judicious decision making
3. Theory of mind

The real dangerous of AI is the bad AI, that build irresponsibly by the corporate. We should not worry on the super smart AI, but on the dumb algorithm.

So if we summarize, the dangerous of AI would be:
1. Bias and discrimination
2. Priority inversion
3. Placing too much trust in automation through machine learning
4. Deep fakes and the upcoming crisis of reality
5. The violation of privacy

His final points are simply take the responsibility of the technologies you create, make them safe and don't oversell it. Remember that the collaboration of human brains with the AI will be better than either of it.

Our first speaker in today's first plenary is Wojtek Krok, Partner at McKinsey. His speech would be more in the business and economic angle of AI, and how McKinsey believes that the world around is changing, also put more positive spin in the whole story.

AI is changing the world we live in, whether we like it or not. McKinsey would like to frame this situation as the fourth industrial revolution. There are five things that fundamentally changing in the world of business.

The first thing is the computing power we have are tremendously increase. Second, people are pouring billions of dollar in start-ups that do AI research and AI products.

According to McKinsey global institute, if we continue to utilize available technologies and apply it to the existing industry, the cost efficiency may reach up to US$3.5 - 6 trillion.

But, as per today, AI is still not the top 3 agenda in the most of business, as it doesn't get the actual business impacts from the technology. People are still struggling to fill the gap between the data science world and the business world.

His last point, if you want to make AI or machine learning work in the organization, you shouldn't focus so much on only the modeling. We need to focus more on how to embed things in our business. Don't only look at the data, but also the intervention you could make and culture.

The two main messages of AI in business are first, the potential is very big, but comes with second, it is very difficult things to do

We continue the session by having Mr. Imron Zuhri from Dattabot Indonesia as our second speaker. He will present his views on Big Data, Blockchain and its applications in emerging economies.

He begins by delivering that the growth of Big Data and AI are all in Asia. The benefits of those are to improve business process and human experience in the region. The biggest challenge is still data integration, as there is no linkages or connections between data.

The biggest surprise in the region is 1.1. billion people exists without a proper ID, out of 7 billion. Low digital penetration in the most emerging economy, for example Java (main business island of Indonesia), the internet coverage is only 30% of the whole area

Then, emerging economies still struggles to equally penetrate the internet across the region. If this is not achieved, then talking about industry 4.0 would be non sense. The first utilization of AI in region is for fraud detection.

On the blockchain side, emerging economies also adapting faster than any part of the world. Even unicorn is paying attention for this blockchain phenomenon, as possible ways to have more business models.

The biggest challenge on blockchain is still on blockchain regulations. But, surprisingly a lot of Asian countries quite open for blockchain.

In Indonesia, we don't have any data on our agriculture. We only have projections that based on another projections. But, asymmetric information is play the biggest role as the cause of low productivity. We solve this by simply distribute ledger to record the transaction.

He close his speech by pointing that algorithm of the computer should help us to trust another human-being more rather than believing computer that might contain more bias

Our final speaker is Jake Lucchi, Head of Content and AI at Google Asia Pacific. He first sets example of Google products that rely most on AI: Google Translate, which rely on machine learning to improve on the accuracy and increase the number of languages they able to introduce.

One of the area where a lot of machine techniques yield is in the health care space. Google has been in partnership with doctor and hospital around the world to see how computer and other types of technology can be used to help diagnosed diseased.

In opportunity side, we think a lot on how we can develop tools or platform that can empower the broader community. To make sure there are more people can take advantages of the advancement of AI.

Machine learning so far is not user friendly, takes a lot of expert to use it. Google cloud is an area when we see how we can expand machine learning to someone who never use that sort of technologies before. Cloud is also more accessible due to its cheap computing power.

The key ingredients for you to able to do machine learning: data, algorithm, and computing power. Our biggest block to do machine learning was computing power.

Last year, we released Google AI principle, consists of seven things we want to achieve with AI technology. It covers the whole of spectrum of things from social benefit, avoiding bias, safety, privacy, and being made available (accessible and able to empower community)

We also have red lines (the things we don't want to do), and we make use of this red lines as the way of analyzing particular product, potential deal we might do with partner or piece of research that we might publish. If anything go beyond the red line, we won't proceed it.

He closes his speech by stating that Google AI principle already get recognition across the world, as multilateral organization like OECD already agreed on Google AI principle. He hopes that further development of AI will only impact positively on humanity.