Boris Cergol is Head of AI at Comtrade Digital Services, a provider of software engineering services and solutions. Prior to joining Comtrade, he was CEO and co-founder of an AI consulting and implementation company and held various research positions in both academia and industry.
Boris currently also serves as executive board member of the initiative AI4Slovenia, is a member of the AI Top Team advising National Standards Authority of Ireland and is one of the mentors at the Techstars Montréal AI Accelerator.
Boris holds a bachelors degree in Theoretical Mathematics and a PhD in Mathematical Statistics, both from the University of Ljubljana.
One of the most important development in AI this year 2020 was the appearance of extremely large language models. The most prominent among them is GPT-3 with 175 billion parameters that was created by OpenAI. The models of this size are capable of generating a huge variety of very convincing texts, from technical documentation to poetry, solve language related problems and hold coherent and sometimes quite interesting conversations. But the most surprising result was the capability of the models to generate working computer code just by giving natural language descriptions of its functionality.
In my talk I will share some impressions of working with the GPT-3 model, demonstrate examples of model generated code and discuss the implications of it for the future of software development practices.