"Achievements in the field of science always allow at
least a small part to touch the future. When forming possible scenarios for the
future of the state, the analysis of the evolution of science can become one of
the most important catalysts for the country's breakthrough. Today, the eyes of
the whole world are focused on artificial intelligence (AI) innovations and
their impact on the future economy and social life. According to AI expert
Dovydas Čeilutkas, this is a field of science that we must pursue now.
D. Čeilutka is the president of the Lithuanian Artificial
Intelligence Association, a data scientist at Vinted UAB, head of the data
science program at Turing College UAB, a member of the State Progress Council,
contributing to the preparation of Lithuania's future vision "Lithuania
2050".
How are the issues of science, AI development and impact on
the state reflected in the current vision of the future of Lithuania? Are they
getting enough attention?
In making clearer steps for the future of the state, we
should pay much more attention to science. Science is a necessary condition for
the successful existence of AI, and AI is extremely closely related to business
breakthroughs, it is a particularly important technology for the economy of the
future.
Unfortunately, we still lack the understanding that we will not achieve
the growth of AI and other advanced technologies if we do not have an orderly
system of education and higher education. We have a lot of catching up to do.
AI is one of the fastest growing fields of science. Things
that were important three years ago are completely obsolete today. Both the
state and business must be able to quickly absorb AI innovations. However, we
are still not able to quickly adopt the innovations carried out in the leading
countries, we lack a central engine. This is holding back and will further hold
back AI projects in the future.
Lithuanian science lags behind in the field of the latest
technologies, achievements related to innovation are poor. Most achievements in
the field of artificial intelligence come from Lithuanians who are self-taught
or have graduated from foreign universities. This does not mean that
Lithuanians will never achieve anything in this field, but most likely they
will have be returning from abroad or creating while there. This is one of the
biggest weaknesses in achieving the desired future scenario of the state. On
the way to it, science should get a central role.
What are the most important drivers of AI innovation today,
do businesses and government policy makers manage to catch up and understand
them?
When a new technology is developed, it often involves people
who have no formal training in the field, e.g. now there are many physicists,
econometricians, mathematicians and even philosophers working in the field of
artificial intelligence. It is important that innovation pioneers who
understand the value of new technology pass on their knowledge and are able to
prove that the new technology can successfully solve problems that have not yet
been explained. However, the activity of technology self-taught people is not a
systemic solution, and systemic solutions are extremely important when forming
the state strategy.
While self-taught people are successfully delving into the
benefits of innovation, familiarizing businesses with the latest technologies
is difficult. Big world companies like Apple or Google solve this by hiring the
smartest and most accomplished people in this research field, working in
world-class AI development laboratories. Research-oriented companies, such as
Open AI or Deep Mind, work extremely closely with academia and contribute to
scientific research activities themselves. AI talents are interested in working
in such companies, because the person doing the research can solve new problems
relevant to the business. Both sides win.
Do we have anything to offer world-class AI talents in
Lithuania? How can we attract them?
We face several obstacles when it comes to attracting
talent. The first is that we are small, it is difficult to compete with, for
example, Holland or Switzerland, which have larger international cities.
The level of wages in Lithuania makes it extremely difficult
to compete with the wages offered in US or Western European companies.
Arriving talents often want to work with the academy,
especially Lithuanians returning after studying or working abroad.
Unfortunately, cooperation with the Lithuanian scientific field is rarely
successful. Lithuanian science is quite closed - this is a big problem.
Some people are also discouraged by the geopolitical
situation. When the events between Ukraine and Russia started, we organized an
international conference, we invited many famous international scientists, but
people were afraid to come, some chose to participate remotely. We have a huge
border with Belarus, which is equivalent to Russia in the eyes of Westerners.
However, bigger Lithuanian start-ups (English scale-up) are
starting to work with problems of the same scale as the big global technology
companies. We have data, an interesting problem, incoming talent can find a new
phenomenon, and we can solve a specific problem. So we start to have traction
points, good data infrastructure, new working technologies, tools needed to analyze
these problems.
The disadvantages I have listed, which make it difficult to
attract science and innovation talent, are not insurmountable, but an incentive
is needed to help attract this talent. Creating these incentives can
successfully counterbalance existing weaknesses.
You mentioned that the cooperation of incoming talents with
the academy does not bring the desired results. In your opinion, what is the
reason for the closed nature of Lithuanian science?
I think there are dozens of reasons, but if one of them is
solved, the scientific situation in Lithuania would improve significantly.
These are economic circumstances and the distribution of funding. Funding of
science in Lithuania is incredibly poor. Our professors earn less than
second-year students - this is information that reaches me from people working
in the academy and official statistics.
Friends, scientists, share examples of job advertisements
that indicate a salary of 300 EUR. These numbers throw you off track, you feel
like you're living in a parallel world. Private institutions can offer ten
times the reward for a scientist's knowledge. So far, all the people I've met
who work in the academy do it purely out of personal motivation, without thinking
about money.
It is strange that there is no real funding for science, but
a lot is invested in the restructuring and construction of buildings. It seems
that academic institutions are primarily managers of real estate rather than
problem-solving engines for the improvement of science.
The growth of science is partially hindered by restrictive
laws and evaluations. Our scientists are not evaluated for their participation
in international conferences, although in the field of AI it is the most
important indicator of a scientist's performance. While scientific journals are
extremely important in many other fields, they are less relevant in AI. The
speed of the AI world is extremely fast, conferences are more suited to the
pace of this field.
Can we keep up at this rate? How does Lithuania currently
look on the global AI map?
We look really sad on the global AI map. In Lithuania, PhD
students or researchers working with AI do not have presentations at the
world's largest conferences, such as NeurIPS, ICML, CVPR, which are like the AI
Olympiad or the Oscars. It is the epicenter where all new ideas and concepts
are presented and discussed.
It's sad, but Lithuania is cut off, the AI world knows
nothing about us. There are many good people working in the field of Lithuanian
science, but they are scattered, there is no ecosystem that helps to grow,
create change, no systemic effect is created, and individual scientists are
unable to change the course of scientific development.
We see success stories abroad, where talented AI researchers
from Lithuania work, we keep in touch with them: Gintarė Karolina Džiugaitė,
Justas Dauparas, to whom we presented an award for merit last year, and dozens
of others.
There are solutions, but they must not be evolutionary, but
revolutionary, radical, otherwise we will hardly be able to change anything. We
are not the only ones who want to be better, competition in this area is
growing between all countries, every country understands that AI is a
particularly important technology that will become the engine of the economy
for the next century.
What is the role of AI in the development ecosystem of
businesses, especially startups? Are Lithuanian startups moving in the right
direction?
We see a lot of good changes in the ecosystem of Lithuanian
startups, especially in the laws and tax base. A decade ago, creating a startup
in Lithuania was extremely difficult. There were no employee options, and their
taxation was extremely unfavorable. Most chose to create startups in Britain or
the United States. However, there is a change, progress has taken place, this
encourages people to choose a more favorable legal base and create a new
startup in Lithuania.
We have two types of startups in Lithuania. Due to their
size, the previously mentioned scale-ups take on interesting problems and are
able to solve them with the help of international talents.
Many smaller startups base their operations on AI
technologies early in their development. Many of our association members are
small startups whose main product is AI solutions. Thus, even a team of just a
few people can work on interesting problems - AI allows solving new global
problems for Europe and the world, rarely limited to Lithuania.
For example,
Pixevia's autonomous store concept developed in Lithuania can be successfully
applied in any country in the world.
An autonomous store is an example of the use of AI in
everyday life. Do we need to awaken the curiosity of the general public - how
can we use AI more widely in life, or is it knowledge that bites everyone?
The saying of Andrew Ng, an AI expert and scientist who has
done a lot for the world, that AI is the new electricity, has already become
famous. Not many people understand exactly how electricity works. Most likely,
only at school did we get acquainted with the basics of operation, understand
the safety requirements. This will be the understanding of AI in the future.
Already, it would be hard to find a product, especially from big tech
companies, that doesn't have AI components. Tens or hundreds of these components
are commonly used. For example, many of us who in the video chat used in the
service, we would count at least ten components of AI technology (models for
object segmentation, audio-to-text translation, etc.), and the team that
created it would reveal several dozen more. But we as users do not need to
understand how the model of the artificial environment in our conversation
works, how it is updated or improved - we have the result and use it for our
daily needs.
In the future, there will be even more user-friendly AI
models, and we will learn to understand them in the same way we learned to use
electrical appliances. The new and currently very relevant ChatGPT is already
known and used by many people. Apparently the first product based entirely on
the use of AI, the general assistant is useful for many people who are not AI
experts. More and more such products will appear, they will be more specialized.
Applying AI in everyday life will be easy and simple if we
have a good education about the main aspects of this innovation, the consumer
side of AI. In Lithuania, a few people are engaged in AI education, but in the
future, if a few hours are devoted to familiarization with the basic principles
of AI, this knowledge will be sufficient for everyday human needs, using AI
products.
What do you see as the global threats to the development of
AI? What arrangements do we need to make so that AI developers and key managers
choose a utopian rather than a dystopian direction for AI growth?
Education will help people understand that there is no need
to fear and hope for the worst case scenario. The aforementioned Andrew Ng says
that thinking about a dystopian AI future is like thinking about overpopulation
on Mars. My attitude is the same. It is true that most surveys of AI experts
show that in the 2050s we will create a general AI (artificial general
intelligence) - a single AI model smarter than a human. Almost all future
scenarios in this context are dystopian, looking at how superAI will take over
the world. But these potential threats are already being addressed by a great
many smart people.
It is worth getting acquainted with the research of Nick
Bostrom, Eliezer Yudkowski and their colleagues. We need to think of AI like
nuclear energy - it's a very useful thing, but we can't build nuclear power
plants and then tackle security. We need to analyze the risks beforehand in
order to make the AI models safe. It's hard to do, but we have the time, the
advanced research teams, and the talent to address these issues right now.
While scientists are taking care of our security in the
future world of AI, what kind of agreements should we as a society and the
state look for in Lithuania right now, aiming for a positive "Northern
Star" scenario rather than a dystopian one?
Looking at the "North Star" scenario, I think we
don't have a mechanism, a model, to help achieve the goals of this direction.
Data science, statistics, machine science always rely on models that reveal
causal relationships and help to plan further steps. We said that we want to
live well, the state must be resilient, the nation motivated, economic
well-being achieved, social security, beautiful nature nurtured, and there is
no shortage of funds in the regions. These are nice goals, but we seem to have
no conceptual mechanisms for how we want to achieve them. We don't have a plan.
I think, moving towards the desired scenario, we must first
develop the economic growth of the country. As the economy improves, we will
get closer to our goals. The role of AI in achieving economic growth will be
crucial. Heads of state and CEOs of major companies around the world already
understand the importance of AI for the world of the future, even though we are
currently witnessing the beginning of the technology's development.
Just like the huge changes that electricity once brought,
the changes of the "new electricity" - AI - will be very significant
in the life of our country. The ability to create technologies and products
based on AI can be the main engine that will allow Lithuania to catch up and
overtake many other countries. However, we need to organize science,
cooperation between science and business, and the promotion of the startup
ecosystem. These are the main elements we should focus on. You don't have to be
distracted by a hundred different things. By pursuing the above-mentioned
goals, we will eventually get closer to the elements of a good life that are
extensively listed in the "North Star" script.
The State Progress Strategy "Lithuania 2050" is
prepared using the innovative method of future insights. The duration of its
implementation is more than twenty years (from 2024 to 2050). It is planned to
present the project to the Seimas in the spring of 2023. The State Progress
Strategy "Lithuania 2050" is prepared by the Government Chancellery
in cooperation with the Center for Strategic Analysis of the Government
(STRATA), the Future Committee of the Seimas and Vilnius University. It is
implemented as part of the project "Establishment of an Evidence-Based
Management Competence Center, No. 10.1.1-ESFA-V-912-01-0025", which is financed
by the European Social Fund. More information at www.Lietuva2050.lt."
Komentarų nėra:
Rašyti komentarą