A talk with Daria Hvizdalova: GoodAI and the future of humankind at BLAST 2017

by | Apr 27, 2017

BLAST 2017 is getting closer: just few days and you’ll meet our great speakers, such as Daria Hvizdalova, Project Coordinator at GoodAI. It will be an unmissable occasion for all the AI enthusiasts as this following talk with her will prove. Book your pass now and don’t miss the chance to attend her and other professionals talks.

GoodAI is a company with international team of 20 researchers is based in Prague, Czech Republic and interested in “building general artificial intelligence, a software program that can learn, be creative, and adapt to the environment in the way a human does”, as it’s possible to read on the official website. On the website is written also: “Our mission is to develop general artificial intelligence – as fast as possible – to help humanity and understand the universe”. GoodAI has several areas of concentration and very ambitious goals, among which the AI Roadmap Institute – a new initiative to compare and study various AI and general AI roadmaps proposed by those working in the field. Could you explain more about it?

Artificial intelligence is a trending concept now, with task-specific AI algorithms being widely researched and developed. Although such algorithms sometimes outperform humans in particular pre-determined tasks, they do not have an ability to extend their capabilities to new domains. At GoodAI, we see that as a significant limitation, and focus on creation of general-purpose artificial intelligence. Due to expanded skillset, general AI will be able to learn and come up with creative solutions for a range of multi-domain tasks.

There is a significant lack of unified approaches to building generally intelligent machines. Artificial intelligence research frequently operates within a specific and clearly defined field of focus, often without considering the importance of the “big picture”. In an attempt to provide a platform for better collaboration and understanding between researchers and to measure genuine progress in the search for general AI, we founded the AI Roadmap Institute.

It is an initiative to investigate how to effectively and meaningfully collate and study various AI roadmaps proposed by those working in the field, map them into a common representation and therefore enable their comparison. The institute strives to devise architecture-agnostic common terminology to compare roadmaps, allowing research groups with different internal terminologies to communicate effectively. The institute is concerned with “big picture” thinking, without unnecessarily focusing on local problems in the search for general AI, and is open for the entire scientific community and everyone keen to contribute.

Let’s talk about General AI Challenge, an international citizen science project launched by GoodAI with $5mil in prizes: it seems a very interesting topic for our audience, full of AI enthusiasts. What about this challenge?

We have launched the General AI Challenge this year to tackle crucial research problems in humanlevel AI development. GoodAI will distribute $5mil in prizes among the various rounds of the multiyear Challenge in partnership with NVIDIA, Microsoft CZ&SK and with the support of many other companies. The Challenge kicks off with a 6-month “warm-up” round, open to the worldwide community of researchers and programmers (both for individuals and teams).

This round is devoted to gradual learning – the ability of AI agent to learn and reuse the skills for problem-solving. We want to incentivize talent to focus on developing safe and beneficial general purpose artificial intelligence. General AI is humanity’s biggest frontier and the ultimate leverage to expedite advancements in science and technology, tackle diseases, and learn more about our universe. The Challenge is based on the “citizen science” approach to research needed to push the boundaries of AI. We are currently working on the upcoming rounds of the Challenge to cover various issues of AI research and development, such as AI safety and futuristic Roadmapping / scenario planning.

A long-term goal of GoodAI is to build a general artificial intelligence software program that will automate cognitive processes in science, technology, business, and other fields. At what point are you, compared to the goals that you have set?

General AI is one of the most complex problems the research has ever faced – and each company is doing it in their own way, thus it is hard to evaluate the current standing point. At GoodAI, we believe we’re on a good track to solving the problem of gradual learning, which we perceive as the foundation of further progress on general AI and one of the most important frontiers. In our research roadmap, we seek a balance between what kind of skills need to be programmed into AI, and which will be learned in what we call School for AI. Gradual learning ability belongs to the programmed part. After that, the focus is expected to be on the teaching and training of the AI agent in the School.

The School is a carefully crafted curriculum of various tasks in artificial environments with increasing complexity. By presenting the right tasks, we as human mentors have a good chance to guide the development of AI and teach it skills and abilities that we find useful. We have an internal deadline of 10 years for the development of general AI – in that sense, we are at the point of having 3422 days left J

What about your path until now? There are some milestones you want to share with us?

We view intelligence as a tool for searching for solutions to problems. The guiding principles of our AI research revolve around an AI which can accumulate skills gradually and in a self-improving manner (where each new skill can be reused and improved in the accumulation of further skills). So the most important milestone we are facing at the moment is to solve the above-mentioned gradual learning.

Some of the core principles of or research have inspired our framework document, which describes how we understand intelligence and provides tools for studying, measuring, and testing various skills and abilities. The framework itself is implementation agnostic, and provides an analytic, systematic, and scalable way to generate hypotheses that we in our team find relevant in the search for general AI. In order to scale up our research process, we started the Challenge and AI Roadmap institute.

Alongside our R&D roadmap, we also have futuristic roadmap. This one deals with societal impacts and safety aspects of AI development. Feel free to read more on our up-to-date progress here: https://www.goodai.com/our-work

How can the AI solutions maximize the business success?

Globalization and digitalization of business has become a widespread trend, so in order to catch up with the new business environment, companies have to invest in the new technologies. However, since AI is a very broad concept, the particular solution is always specific and tailored to a specific need. We do that at our sister company, GoodAI Applied – it uses GoodAI prototypes and state of the art artificial intelligence solutions to maximize business success for companies and organizations across a range of industries.

We analyze how AI can help the business, do the necessary research, and implement practical solutions. GoodAI Applied started because of high demand for the know-how and experience of GoodAI researchers. Our R&D team focuses solely on general AI research, and is building technology that won’t be on the market for another 5-10 years. Clients of GoodAI Applied receive exclusive access to the GoodAI core team and research results. We always quantify how much value we can bring to our client at the start of any project. We limit risk by working with minimum viable prototypes, and test project viability early and often.

What is the current situation of the AI industry in Europe, compare to the rest of the world and, in particular, to the USA?

AI industry is very specific and wide – nowadays it is basically perceived as an extension of data science, and comprises both general and narrow AI technologies, with over 99{c653bac874e230c60cb5d5d5e780b44e2af54b021d2579c749fcb3ad96398bd8} being narrow AI research and implementation. The whole world is booming with AI start-ups for narrow AI, and it does not seem to me that geography implies any borders in that process. There is only a handful of private companies focusing on general AI, the potential of which is widely acknowledged – and we see them getting more investment both in Europe and overseas.

In your opinion, how should we interpret the interest of several giants as Elon Musk or Mark Zuckerberg (or Google as well, just to name a few) for AI?

As a statement of AI’s importance for digital and technological businesses! Their willingness to invest in the development of AI, especially in-house, is a promise that AI is an enabling technology for any other field, and curious things are about to happen in the next decade.

You’ll be one of the BLAST speakers and you’ll talk about “Accelerating Towards General Artificial Intelligence: GoodAI and the Future of Humankind”. Could you anticipate us something about?

General AI is quite a complex concept – I would like to break that down for the AI enthusiasts and elaborate on our vision and current activities.

Thanking you for having joined BLAST, what convinced to believe in this adventure?

I am excited to join the discussion on future paths of technological progress, meet inspirational speakers from all over the world and share the knowledge we have at GoodAI!

[fbcomments]
Share This