Learn how the World Bank Group is helping countries with COVID-19 (coronavirus). Find Out

Speeches & Transcripts September 25, 2019

Making AI Work for Ho Chi Minh City

Ho Chi Minh City, September 25, 2019

Excellency Mr. Nguyen Thien Nhan, Politburo Member, Party Secretary of Ho Chi Minh City

Mr. Nguyen Thanh Phong, Chairman, Ho Chi Minh City People’s Committee

Leaders of Ho Chi Minh City departments and agencies

Leaders of universities, research institutions, private sectors,

International speakers, ladies and gentlemen,

 

Very good morning – xin chào!

I had the opportunity just last week to visit and discuss with a number of leading Vietnam private sector corporations the innovation, science and technology agenda, as a part of the country development strategy in the next ten years. There is no doubt that the Artificial Intelligence (AI) revolution is already affecting the way most of us live, work, and play.

The fact that HCMC is hosting this conference today means that its leaders understand this. So the question is not about the “why”, but is really about the “what” and the “how”. What and how specifically can AI deliver results for a more prosperous and resilient future of HCMC?

Let us start with the “what”.

The fundamental challenges that HCMC faces towards today are not new.

But rapid growth is amplifying concerns about traffic, floods, environmental quality, and infrastructure gaps, and we should all agree to the need for providing good jobs, better public services and better quality of life. Massive infrastructure needs, but significant fiscal constraints, mean that HCMC government needs to be smart about urban planning, including for land use, where to build new infrastructure, and ensure existing infrastructure delivers, whether for transport, health, education, water, waste management, IT, or flood risk.

At the same time, HCMC is seeking to improve its public services and decision making, as well as attract private investment.  Making HCMC a preferred destination of private investment, including for digital economy and innovation hub in the region, and even globally, means making the city attractive and livable for talent.  It also means providing local innovative enterprises with access to business opportunities and data to solve real world problems.

AI can offer solutions to some of these challenges, given the right enabling environment. Companies and cities that are smart about AI will reap the rewards.  

The next question is the “how”.

AI mimics how the human brain works. The explosion of digital “big data”, exponential growth of computing power including from cloud platforms, and massive innovations and investment in AI methods and algorithms means that AI can serve to do certain things quite well!

Google Maps and Grab Taxi uses AI to find optimal traffic routes. Argentina’s biggest city, Buenos Aires, uses AI to target optimal maintenance of storm drains. New York uses AI to better target food and safety inspections. These are only three examples, and today you will learn about many of the rich and very practical use case in how AI is being used to address real world problems and opportunities. Global hubs as Shanghai, Singapore or Seoul have been mainstreaming AI into all aspects of their city’s DNA!

But AI is ultimately a tool that must serve humans. Therefore, let me stress three key elements for AI’s success for HCMC:

·         First, setting clear and realistic expectations for where and how AI can deliver for HCMC.

·         Second, ensure that there is an enabling environment for AI to succeed in practice, especially when it comes to accessing and integrating the data needed to solve the city’s challenges

·         Finally, make sure that we understand and manage any key risks associated with AI.

A.           Successful AI Requires Setting Priorities Beyond Business As Usual

Applied AI broadly does two things: automate existing human process (including eventual driving!) or open up whole new realms of services and decision support.

To paraphrase Apple founder Steve Jobs, people don’t always know what products can transform their lives. He joked that if early car makers had asked people with carriages what they wanted, they would have said faster horses!

So, let me suggest that successful AI cannot just be about automating business as usual. In other words, IR 4.0 cannot run in the bureaucracy 1.0! Opening up new opportunities will require ensuing that people and processes -- including the enabling national and local legal and regulatory framework – progressively embrace and align to the emerging world of AI applications. 

Especially in government, computerization and now AI must avoid simply trying to automate exiting business processes. Successful digitization provides the opportunity to reform, rethink and streamline processes.  HCMC’s challenges require working across government agencies, the private sector, and academia. AI would ideally transform what HCMC can realize in terms of data-driven decision making and service delivery, while promoting a vibrant and innovative private sector.

B.           Real AI Requires Enough Data, the New Oil!

AI is likely to have limited real-world relevance if it cannot draw on adequate digital data, suited for the objective at hand.  While AI can yield important results, success depends on aligning actual digital data to a particular problem or opportunity. This also implies the need for being aware of the possible risks of not managing AI and its underpinning digital data well.

The global value of digital data now appears to have surpassed that of oil as an asset! Clearly, data has become the new fuel of our economies and societies. But just as gasoline has little benefit unless used to power cars, machines or trucks, AI depends on adequate data!  AI runs on the right digital data. No computer AI will help address a real-world problem if it is expected to perform with no fuel or sense of direction! If our brains need information to think, an AI system needs digital data to function!

AI data sources can include existing systems (like infrastructure registries, tax records, administrative data), but also richer sources such as sensors, smartphones, or satellites.  The ultimate value of these is not in the hardware, but in the data and insights these provide, especially though AI.  But even Terabytes of satellite or sensor data will have little value if they cannot be access or distilled into actual insights.

For HCMC’s key challenges, data assets for AI could be derived from international, national, or local platforms as sources. Today’s speakers will give you a good sense where HCMC already has assets in its reach, but also where it is falling short.

While technology issues in terms of access, storing and processing big data for AI are important, the biggest challenges to succeeding in this aspect of AI remain people and institutions.

Data sharing and integration does not come natural in many governments.  And while governments with increased digitization are increasingly a treasure trove of data, there treasures are typically lost or fragmented across many different sources. Common data sharing platforms, like HCMC is implementing, are a start to addressing this problem.  Where data can be put in the public domain, open data can also serve as fuel for local innovation, as well as government itself making better use of its data. 

With strong leadership, the drive for successful AI however can be a strong catalyst towards getting key data to be actually consolidated, used, and generating value. Cloud based infrastructures now provides for strong platforms for data integration, and security, as well as linked compute power to deliver on AI.  Cloud architectures are better than having data disbursed about a whole host of stand-alone and disconnected traditional computer setups.

C.           Smart AI Application Should Maximize Rewards, But Also Manage Risks

If applied well, AI promises to have significant and tangible benefits for HCMC. AI can help the government make better decisions, as well as offer services that are more responsive. Responsible data sharing with the private sector can also stimulate the local economy and make consumers better off.  But AI with big data that is poorly thought through or applied can also have significant risks.

Key risks are the hype fails to meet expectations and that an abstract pursuit of AI and big data is an expensive distraction in terms of time and addressing real issues.  At the time, in our digital society we need to ensure privacy and cybersecurity well managed, and that the decision made by AI are well understood and not subject to undue biases.  As AI does certain tasks better, we must ensure that all HCMC’s officials and citizens build the skills and digital literacy to thrive this new age. This will be critical, because no matter how smart AI is, it is ultimately controlled by us to solve our own problems.

We hope that you will use the next two days productively.  Today will provide the opportunity to be inspired by what AI can now do, and also to learn where some of its pitfalls lie.  Tomorrow, a wider set of officials will take a series of deeper learning exercises to relate AI and the “must-have” data to address some of HCMC’s key challenges.

In conclusion, seizing AI’s opportunities in a result-oriented and disciplined manner holds significant promises for HCMC.  But the failure to adequately resource and manage key cross-cutting initiatives will yield disappointments. Terms like Smart City, Industry 4.0, Digital Economy, and AI must translate to something concrete and ideally positive for peoples’ lives.

As the World Bank, we stand ready to continue moving forward in this journey with you. HCMC’s pioneering efforts in digital government and AI provide a way to feedback to the national legal and regulatory enabling environment that often shapes what provinces are able to do. But at the same time, cities and provinces, like HCMC, can be the real incubators for the successful application of disruptive technologies like AI.

I look forward to our very productive and fruitful discussion.

Xin cảm ơn!

Api
Api