Forum

HLEG - Input Request: Business Impact

Ieva Martinkenaite

15 November 2018 - updated 2 years ago
7 comments
Total votes: 2

Dear European AI alliance members,

Business Impact subgroup of AI HLEG is working on delivering recommendations to EC on how to ensure the uptake and acceleration of AI by European business.

Below is our proposed structure for delivery, based on 4 assumptions and fact-based knowledge:

75% of European business does not apply AI;
25% of European business apply AI but struggle to accelerate and scale;
Europe is behind other world powers in terms of home-grown AI unicorns and tech giants;
AI business uptake and acceleration is enabled by a dynamic interplay of ecosystem components.

We seek YOUR INPUT on the following 3 questions:

What can European Commission (or Member States) do to enable uptake and acceleration of AI in European business? What are the priorities?
What are the key components of enabling AI ecosystems? Can you give examples from industries, particularly, automotive, health, industrial machinery and finance?
Are unicorns and tech companies important, and in what way?

Any other input on delivery structure or else on AI uptake in business across Europe is also welcome!

Thank you in advance!

Ieva & Loubna (co-rapporteurs, AI HLEG, Business Impact subgroup)

2.1. BUSINESS IMPACT - DELIVERY STRUCTURE

INTRODUCTION (focus: ambition level for Europe)
SECTION 1: CONCEPTUAL FRAMEWORK (list of enablers for AI adopters, AI accelerators, Europe-born tech companies & unicorns, AI ecosystems)
SECTION 2: AI READINESS IN EUROPE (as-is situation in comparison to USA, China, Canada, etc. across 4 dimensions, ref. Section 1)
SECTION 3: AI UPTAKE DEEP DIVE (enablers & recommendations)
SECTION 4: AI ACCELERATION DEEP DIVE (enablers & recommendations)
SECTION 5: TECH COMPANIES and UNICORNS (enablers & recommendations)
SECTION 6: AI ECOSYSTEM DEEP DIVE (health ecosystem as use case; automotive / finance / industrial machinery selected for future deep-dive)
SECTION 7: RECOMMENDED ACTIONS (what and how, by whom, at which authority level, e.g., EU, Member States, regions)
CONCLUSION (focus: summary, networks of networks, KPIs to check progress)

Tags:

Comments

Keith Tayler,

16/02/2019 17:22

Has the Business Impact subgroup of AI HLEG done any research into whether so-called 'AI' computer systems increase productivity, profits and GDP, or it just another 'assumption' that it does? We must not forget that over the last 25 years there has been no approachable greater increase in these three economic measures. Globally there has been an above average increase, but these have been due to political and economic change and cannot be attributed to increased technological changes. Indeed, given the 2000 dot com recession and the 2008 deep recession we are still in, which were for the most part casued by over hyped and misused IT, the usual increase in these measures due to technological improvements since 1800 would if anything appear to have decreased since the 1990s.

Without any evidence that ‘AI’ technology is or will benefit European business and citizens, it is surely irresponsible to bring pressure upon or incentivise businesses to adopt ‘AI’ technology. For sure, business should be encouraged to make informed decisions about new technologies, especially when much of it is untried and is for the most part still at the R&D stage. This appears to be just another example of ‘AI hype’ and does not demonstrate much human intelligence.

Michael Clarke,

31/01/2019 08:09

What are you expecting them to do with it? AI for AIs sake is a waste of money. How can these companies deploy it such that it will given them a positive return on their investment.

A lot of AI these days happens in 'the cloud'. There are a lot of legal ambiguities around the cloud, especially when the data ends up being hosted on a cloud in a different country and the Ai gets applied to it in a third while the customer interaction is happening in a fourth. A bit of clarity, a legal definition of exactly where the transaction is occurring and some requirements on the cloud providers to properly manage the data could help, especially for companies in the smaller/less advanced EU countries.

Lluis Esquerda ...,

10/12/2018 17:53

Hello Ieva,

Thank you for asking for inputs on this topic as it will be one of the most important.

I am going to reply to the three questions you presented. But firstly, I will reply to the second question.

First I will respond the second question because I think that before we propose actions to tackle down a problem, we must understand the problem and have an eagle sight perspective to see all the components of the ecosystem.

Beginning with the first question, and going into the most simplistic key components that enable AI in businesses (not going into engineering issues like communication standards), we could divide them into three subgroups:

1. Business enablers

2. “Environment” enablers

3. Consumer enablers

In the first Sub-group the Business enablers, there are at least 3 key components:

1.1.Benefits

This is probably the most important component in the ecosystem to enable AI. If there are no benefits to use such a technology, a Chief of a family business is not going to invest in it. Because the common thought is that you could use that money for other things or even to be richer. That’s how family businesses and SME use to think. So, only by telling them how much they could win, they are going to use it. We will enter to discuss that a bit deeper in the second sub-group.

1.2.Devices and infrastructure

1.2.1 Devices:

Any business needs devices to run the AI software. Some of the devices or machinery have the AI pre-installed in their systems, but others don’t. For instance, a CEO can decide to buy a robot arm that can do a so specific task in the production chain automatically only with the supervision of a human. Like you do in production control. But this type of machines can be quite expensive and do not represent the majority of cases in Europe. The reality is that in most of the cases the AI could be applied to administration tasks like accounting, payments, communication, failures forecast and, supervision of tasks and numbers. In such cases the AI does not require a specific machine with an AI pre-installed, you only need to install an AI software into the company’s computers or machinery.

1.2.2. Infrastructure:

In this case, I do not refer to the to the tech infrastructure of the company, but to the organizational and managerial infrastructure that have to let this AIs work smoothly and correctly. I do not say independently (now), but in the near future, they will.

More concretely, and coming back to the managerial infrastructure, companies have to change the way businesses work. Otherwise, if you give them the most advanced AI in the world but you do not let them work it’s like putting a 600-horsepower electric engine into a Ford-T. Literally useless.

So, the conclusion is that companies must have an infrastructure or organization in the company that enables to obtain and use the benefits of such a technology.

1.3.Peace of mind

When I mention peace of mind, it’s about the comfortability of letting an AI software telling you what is the best option. It looks like easy for a lot of young people but, the reality in Europe is that most of the businesses are managed by people not feeling quite comfortable at the time that a software is telling them what to do with their business or in other words, their money. The statistics from the Eurostat say that in 2015 the 99% of the enterprises in Europe where SME and 94% independent. Here, we find the importance of such type of companies. So, making them (people with the power of decision) feel comfortable with such type of technology it is not something you can change with three marketing campaigns. For instance, I would like to mention two regions that are the contrary, one it is Europe and the other is China.

If you go to China, you will see that in the developed cities, you can ONLY pay with WeChat or another MOBILE application. This is something quite advanced, to be honest, but most important, useful. Chinese people no longer have to take care of taking money with them but, to take care of the battery.

In the other side, we have a capital like Berlin, earlier this year, I was there with no physical money because I live in Barcelona and here, you can pay almost everywhere with your contactless credit card. So, I was there and commerces told me “Sorry we do not accept credit cards”. That happened in most of the restaurants and non-big supermarkets.

So, we are trying to implement AI in Europe when in some countries they do not even accept credit cards. And, credit cards were created and implemented in other countries many years ago. So, we must understand what is the social mentality or the reason because one of the most advanced countries of the EU has not still applied the credit card technology.

A study about this not only in Germany but in every country have to be done in order to apply the necessary actions. This will be further discussed in question 2.

In the second Sub-group the Environment enabler, there are at least 3 key components:

2.1 Regulatory and Policy framework

The regulatory and policy framework (law and rules) is probably one of the most important key components to enable the adoption and full usage of AI. The tough question to answer from the business managerial perspective is “Will the executives be responsible for a mistake done by the AI software?” And, in the case, the Executives are not responsible, then, “who?”

I would like to remember the story of the 2010 Flash Crash (https://www.theguardian.com/business/2015/apr/22/2010-flash-crash-new-york-stock-exchange-unfolded) In that particular occasion, they recovered the prices before the flash-crash and “all” was solved. But, in a complex business ecosystem like the one of the advanced European market, you cannot go back on time and say to other companies to cancel the sell your AI accepted. Or, who is going to pay for the products incorrectly built by your AI?

We should define the way of acting in each one of the scenarios. But as this is a smart technology you cannot predict each one of the scenarios. So, we must think in the general ways of acting. Also, I would like to remember that this is technology, and this type of technology is built by thousands of different software makers. And, not all of them use the same security procedures in its building process. Therefore, the question is not if the AI software is going to fail but, when will it fail and how businesses are going to proceed to solve such situations.

2.2. Tech infrastructure

This is something that will only happen sometimes but, in some cases, you will need third technologies to enable the adoption and application of AI. I am not referring to the technology of your AI provider but, to third general technologies like 4G & 5G networks to enable the application of such AI services.

2.3. Promotion

Regarding the promotion, I believe that without promotion, business people with the power to make decisions will not be aware of all the benefits. More concretely, the promotion should be done by both private companies and public institutions.

From the private side or companies, they need to sell their products, they are winning money, how the EU could enable such a thing, will be discussed in the second question.

By the public side, the EU and other public institutions like universities, R&D centers, and high schools, could have such a key paper. This could be even done by kids showing their parents how to be confident with the latest technology applications. Like we do it today with Siri, Google Assistant or Amazon Echo.

In the third Sub-group, the consumers' enablers, there are at least 3 more:

3.1. Comfort, confidence and convenience

Comfort, confidence and convenience will be necessary only in some cases. These are the ones when users know they are receiving a product or service built or partially offered by an AI. In such cases like a smart band informing you about your imminent heart attack or a self-driving taxi car. In cases like self-driving cars, the comfort, confidence and convenience must be reached by a bit of experience and a lot of promotion. The public promotion will have a key paper.

3.2. Devices

In some cases, users will have to own a device. Like a smartphone or smartwatch to complete a payment or to show a QR code.

3.3. Security of their data

Regarding security and data protection. It’s quite sensitive in some cases like medical records needed to prevent some illnesses or to follow continuously your status.

Here, I would like to recommend an article that appeared two weeks ago about one of the products of DeepMind, company owned by Google. They said to their users that they wouldn’t connect in any way the patients’ medical records data with Google. Now, with the excuse that they have to grow globally, the product is considered as a Google’s product. Therefore, the data is connected to Google. (https://www.theguardian.com/technology/2018/nov/14/google-betrays-patient-trust-deepmind-healthcare-move) We could have such an interesting discussion about that.

Regarding the first question proposed by Ieva, the one that said: What can the European Commission (or Member States) do to enable uptake and acceleration of AI in European business? What are the priorities?

I am going to answer each one of these questions by the order that has been presented in the first question that I have answered (the second presented by Ieva).

The first one it’s the actions regarding the business enablers.

1.1. Benefits:

There are:

1.1.1. R&D projects funded with European funds. The projects could be created by the EU. Like the Eurofighter. Or, could also be created by private companies or students that participate in official European competitions. This way of acting would be more similar to the way of acting in China. But they are many years ahead of us and the rest of the world.

1.1.2. Fiscal deduction by the acquisition of products built upon AI or that has AI as its core. Then, it should be defined quite clearly what it means to have AI on its core.

1.1.3. Purchase of products that use AI on its core and that will enable some type of benefits for companies. In this way, if for example, the EC organises a competition about the best products that uses AI regarding some specific tasks, then, you could promote the winners of such competition and not promoting the products by itself. I say that because we must remember that if the EC or EU promote some specific products but not others, then, you are affecting the free commerce. And equality of opportunities.

1.2. Devices and Infrastructure

1.2.1. Devices

Encourage the acquisition of new devices and machinery that have AI as the core technology. Applying as a well fiscal deduction and a partial funding of such investments in new devices and machinery.

1.2.2. Infrastructure

The creation of some authorities or organizations to advise companies on how to implement such changes in their organizations with the aim of using the real potential of AI or instruments that use AI.

Here, the promotion will have a key paper but, the EC could help businesses with that piece of mind if EC shows that we are already using AI and it really works. In the end, this is a promotion strategy but, I think that it could help.

2.1. Regulatory and Policy framework

The creation of a Regulatory and Policy framework that defines how to act I general ways. I say general ways because as this is technology, you cannot preview all the possible scenarios. To put an example, it is more or less “the same” in self-driving cars. As you cannot tell the AI how to behave in each scenario because it is impossible to compute all the possible scenarios, what we do, it is to tell to the self-driving software how to generally behave.

I would like to put the example of the AI Regulatory and Policy framework presented by the Monetary Authority of Singapore http://www.mas.gov.sg/~/media/resource/news_room/press_releases/2018/Annex%20A%20Summary%20of%20the%20FEAT%20Principles.pdf

2.2. Tech Infrastructure

Create a commission to study and determine the key technologies that will enable the adoption and usage of AI at the European market. Then, once such technologies have been identified correctly and assessed how they should be improved, the EU could:

1. Invest in such technologies. Owning them in this way.

2. Partially fund the private ones.

I am quite aware of what it means for the EU to own infrastructures. But, here, I think that we should compare our actual situation with the one of the US and China.

On the American side, they are more developed than us. But I read and was told that actually, you have to pay for almost everything. That takes us to three possible conclusions: 1. The US government collects a lot of taxes from everything; 2. Most of these places where you have to pay are privately owned so, you pay for having access to it because they invested in those infrastructures; 3. A combination of the two previous.

On the Asian side, as we know the People's Republic of China it is a Communist State. In the Chinese case, this is the country in the world that the Gov is investing more money. So, they own it. Here, I paste the link of the McKinsey Report (https://www.mckinsey.com/featured-insights/winning-in-emerging-markets/chinese-infrastructure-the-big-picture)

This will be one of the most important and key factors in the uptake and acceleration of AI in European business. That happens because we (almost everyone that use Internet services) are already using AI. And, actually, we have it in every popular app or software that we use. But we do not know that we are using it or they do not say it.

So, the promotion could be done in many different ways. From showing to the EU citizens and foreigners that AI is already running in our system to promoting products from the EU that use AI in its core.

So, the promotion must be done by both Public and Private sides.

The actions in the public side could be something like the Government of Estonia who has publicly announced that they are using blockchain in the country national elections.

In the private side, it could be a grant in advertainment by the EU to some private companies who sell or use products with AI in its core. This could be seen by some private companies as a reason to develop or sell AI products.

There are a lot of actions, the EU and the EC could apply in this field.

Regarding the actions that could enable users to have that comfortability, confidence and convenience. The actions the EC could do to enable such feelings are actions of promotion. Something like the States Members do today to try to reduce car accidents or like NGO sell ideas. I feel bad for saying it, but it’s marketing in the end.

I would like to put, the example of China. China is the most populated country in the world that applies AI generally. And, they are quite likely, the most advanced in the world. So, we could ask ourselves, what did they do to enable such a thing and without saying it out loud? Yes, we would agree that there are quite a lot of values that the EU maintains that maybe had not been kept there. I would agree as well. But, as I am concerned, the application of AI in China (most of the cities) it’s generally. And, basically, they can (have to) pay everywhere with the app developed and support by the Government (WeChat or another supported by the Government). In the end, if I am not mistaking, they can pay with four different payments apps. But, as WeChat is also a messaging, search engine, social media and dating app, this is the one most used.

The conclusion is that through the support of the Government, most of the population is using them and enjoying the benefits of AI in payments. So, let’s try to do an action that enables EU citizens and foreigners enjoy the benefits of such technology.

How? I would like to propose a type of annual competition organized by the EU with all the State Members, where the best products in each category compete to be selected as one of the best products in that specific area.

For instance, in the case of mobile payments, the EU organised the competition, chooses the best apps and the winnerS (not individuals) are named as the best apps of the EU for mobile payments.

That support from the EU to some specific areas could help to have the comfort, confidence, and convenience by the users.

We could complain that if the EU organises the competition, then, it could break the equality that the European Market offers. Then, the EU could emphasize (pay or subsidise) one of the most prestigious tech consultancies brands in the world or directly TechCrunch, Forbes or any other media to organise such a competition.

In this way, private companies could use that Technology Products guide to decide what products they should use. With the final consequence of people enjoying the benefits of using products with AI.

Regarding the devices, the EC could help by subsiding some products that use AI in its core. This is done with the aim of enabling people to buy those products and enjoy the benefits of AI. But, we must take care that companies do not use this grant to have higher profits. So, if we can take care of that, I think that it could be quite useful the reduction of price in products that use AI.

The security is also part of that peace-of-mind that will enable more consumers to use products/services that use AI.

Fortunately, in Europe, we already have the GDPR. So, I think that problem, could already be solved. But, let’s discuss it.

Regarding the third question: 3. Are unicorns and tech companies important, and in what way?

I have not developed so much this question because I do think that they key paper it is already demonstrated it in the most advanced regions in the world (technologically speaking). Just take a look at the most advanced regions in the world outside EU (USA, Republic of Korea, Japan, China), they all have tech enormous tech firms and media channels like Google, Apple, Microsoft, Uber, Lyft, Sequoia, Samsung, LG, Sony, Softbank, Tencent, Alibaba, Baidu, etc).

The EC must enable the creation of more tech firm and unicorns in the EU market.

I do have the original Word, I could share it. All is better organised.

Thank you,

Lluis

Roger UPFOLD,

03/12/2018 09:53

What can European Commission (or Member States) do to enable uptake and acceleration of AI in European business? What are the priorities?

As a member of the Machinery Working Group and Machinery ADCO groups concerned with the Machinery Directive 2006/42/EC which deals with the legal framework for the placing of new products in scope on to the single European market (or if not placed on the market, when first put into service, eg in situ manufacturer), and not just 'machinery', but also 'safety components', many of which will be 'logic units to ensure safety functions' (as defined), it is vital that the legal framework is sympathetically adjusted and developed to accommodate the legal and ethical consequences of AI (eg machine learning) in a way that maximises the benefits whilst protecting the interests of those who use or encounter products that use AI as part of their control systems. This is especially important where it may impact on the delivery of health and safety, following the long-standing objectives of the Essential Health & Safety Requirements (EHSRs), including the fundamental hierarchical framework in EHSR 1.1.2 on the 'principles of safety integration': 1 Avoid/Reduce risk by Design, 2 Provide Protection Measures, 3 Warn of Residual Risk, all to the 'state of the art'.

Shalom BROYER,

25/11/2018 10:35

I want to refer to the first qustion.

SMBs (Small Medium Businesses) adopt AI technologies very slow because the result of such a project is not guaranteed. In this case, the managers of the SMBs do not run to finance such projects, and therefore it is difficult to harness them to this issue. Our company (Maxsoft) has received public funding for an attempt to predict outputs in avocado. We have received permission from one of the customers (Avocado packing house) to use data sets of output in the past 20 years. Right now, we are dealing with the subject. But even though the funding still exists, the challenge is to explain the customer the benefits of such research.

If we would be able to disseminate the benefits of such research to these SMBs, they will look for solutions in this area.

Kai Salmela,

22/11/2018 23:51

Thank you for your question. I'd like to answer to them individually.

As I’m working on Finnish DIH, following is my experience:

For a normal company , AI experts are needed for recognising the needs, enabling environment for an AI and bringing up the system. Then there’s a bit of learning curve for the leaders and staff of the company. For this need DIH’s are suitable and EU Commission has already established them. Naturally DIH cannot bring up all the companies, but for this we have a network of companies that we can pair with the customer. Job gets done and DIH just run the project thru it’s cycle.

How to enable uptake and accelerate business with the AI? We really need the European standardization of an AI. Currently only big AI houses ( in the US ) are safe bet for a company CEO. Any other choice, and you may have your board of owners coming to you. AI according EU standard would be another safe bet for CEO, and much better for entire Europe. Standardization would give us also modular AI, where manufacturers could make only those parts that they’re best at. That would bring the cost of an AI down and user could afford to pick up those modules they need and what they know are best suited for them. That offcourse would need a standard between modules for communication. Eu would like AI to be safe for humans and function within laws – there standard modules for law, feedback and control are needed. For example GDPR is one of the laws that every AI should be able to obey – now it’s just not the time for it yet and that is one big obstackle for a company to adopt AI technology. Money just cannot buy everything and we need to mature AI marked in Europe with a standard.

For a priority, EU could make it easier for DIH’s to finance their job. At the moment funding goes thru governmental evaluation with political views and that doesn’t always promote help for small or middle sized companies nor co-operation with the cross-border DIH’s. Is direct funding impossible with importance and achievement evaluation?

What are the key components of enabling AI ecosystems? Can you give examples from industries, particularly, automotive, health, industrial machinery and finance?

The ecosystem grows when there’s a mutual benefit with the DIH and Company that is involved. DIH is non profit company, publicly owned and specialists trusted by default. Company can get free research for their needs and a new partner that can help them in that job. DIH can oversee that process and make sure that company get their need fulfilled and providing company gets adequate compensation. When DIH has education ( like an university etc) to offer there will be more persons that can their way around AI in the companies and society in general.

As we have very vibrant technology cluster in our area, where most of the ecosystem companies and new start up’s are growing, we have’nt forgotten our healthcare either. Common companies get their production automated and robotization is spreading in very good rate, but also healthcare has had it’s own AI innovations and robots for the special need individuals.

One of the most interesting innovation has been our “Living Lab”, that is situated in some ecosystem company where they have opened their facility and data for experiments to another companies. With these labs you can come and test your ideas and find the best ways to adopt automation, robotics and AI in your company. Usually these companies also find some common ground and start co-operation in that process. It is really a win-win situation and our DIH is overseeing it.

Are unicorns and tech companies important, and in what way?

Companies that strike it thru are needed and they start to build an ecosystem of their own around them. That is all beneficial and good for the local economy. However, public funding seem to seek also these companies and when it does , smaller companies are “forgotten” behind. For a politician it is beneficial to direct money for a big company instead of smaller one. We rather would see much more diverse funding than today.

I hope You find these answers any help to Your task.

wbr. Kai Salmela , AI Specialist , Prizztech Ltd, RoboCoast R&D

Ieva Martinkenaite,

23/11/2018 11:56

Dear Kai,

Thanks for your valuable inputs. Testing and experimentation environments, such as those enabled by DIHs are of critical importance to AI uptake in business.

Let me bring your considerations to the group.

Great to see your good work in this area.

On behalf of AI HLEG,

Ieva

European AI Alliance

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Forum

HLEG - Input Request: Business Impact

Comments

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Share this:

Comments