Navigation and service

AI specialists and their skills are very much in demand. It is therefore essential to understand the situation of employees with AI skills in the labour market. How many people in Germany can program, use and control AI? How does Germany rank internationally in terms of its number of AI specialists? Which AI skills are currently particularly in demand – expertise in machine learning, deep learning or another area? The indicators shown in the “Skills” thematic area provide initial answers to these questions.

Machine learning: the undisputed number one AI skill in Germany

Which AI skills are particularly sought-after and relevant in Germany? Two indicators are used to answer this question. The first indicator shows which skills were the most frequently added AI skills among LinkedIn users in Germany during the years 2015-2020. The second indicator reveals the AI skills for which the frequency of mention by LinkedIn users (in the annual comparison) showed the highest percentage increase during the same period.

Among the most common AI skills (first indicator), machine learning has been the undisputed number one since 2015. However, artificial intelligence (see the notes regarding the methodology) and deep learning are to be found among the top-ranked positions.

Turning to the greatest changes in mention of AI skills in the annual comparison (second indicator), what is striking is that both the sequence as well as the skills change utterly almost every year. Whereas the three most important AI skills in 2019 were pattern recognition, language recognition and convolutional neural networks, they were data structures, Pandas (software) and Julia (programming language) in 2020. This indicator paints a clear picture of which skills are behind the commonly used term “AI skills” – especially in Germany.

Top AI skills by frequency (2015-2020) (Germany)

AI skills that were most frequently added by LinkedIn users in Germany during the years 2015-2020

  1. Machine Learning
  2. Artificial Intelligence (AI)
  3. Deep Learning
  4. Computer Vision
  5. Image Processing
  6. TensorFlow
  7. Natural Language Processing (NLP)
  8. OpenCV
  9. Keras
  10. Neural Networks

LinkedIn Economic Graph, December 2020

Note: LinkedIn data does not represent a randomised sample of the working population of a country. It is based on statements by LinkedIn users and as such does not necessarily represent all economically active persons in a country, even if a high proportion of those in many countries are LinkedIn users. The analyses presented here show the world from the perspective of LinkedIn, i.e. results are influenced by who uses LinkedIn and how. Depending on the country, this may be different and is influenced by various factors such as professional, socio-economic, regional or cultural background. In Germany, less than 40% of the working population are on LinkedIn.

AI definition: The definition of AI skills presented here covers a list of skills that were identified by LinkedIn as AI skills. This list includes machine learning, natural language processing, data structures, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, Pandas (software) and OpenCV. The complete list of AI skills is the property of LinkedIn and is not publicly available. For more information about the LinkedIn Skill Group Classification, see also: World Bank Group, LinkedIn (2018), Data Insights: Jobs, Skills and Migration Trends Methodology & Validation Results, Appendix F.

Methodology: Top AI skills by frequency: The data underpinning this indicator relates to the ten most frequently mentioned AI skills on average by LinkedIn users in Germany in the years 2015-2020. Artificial intelligence is also one of the second most frequently added skills by LinkedIn users to their profile between 2015 and 2020. Even though this skill falls under the umbrella term of the AI competence group, it exists as a separate skill at LinkedIn because LinkedIn users specify it as such on their profiles. LinkedIn’s skills taxonomy is derived from the skills that LinkedIn users specify. These are audited by LinkedIn, verified as actual skills and then standardised, which means that the skills taxonomy is largely generated by users. Top AI skills based on the highest year-on-year change: the ten skills that had the highest percentage growth of LinkedIn users in Germany who added these skills to their profiles during the years in question. Further information can be found in the methodological note of the OECD AI Policy Observatory (OECD.AI).

Only 21% of AI talents in Germany are women

AI experts are highly sought-after on the labour market. Companies, universities and research institutions face major challenges in recruiting qualified employees to fill vacancies in the field of AI. How many AI talents are there in Germany and other countries? And how are women represented in this forward-looking area? In which industries do AI talents work in Germany?

This indicator shows the number of AI talents in Germany, the USA, the United Kingdom, France, Switzerland, the Netherlands and Denmark. In Germany, a total of 143,000 LinkedIn users were identified as AI talents. The absolute number of AI talents in the countries cannot be compared directly because the number of those economically active in the countries is different, the proportion of LinkedIn users out of the total working population varies and because industries and labour market areas are represented on LinkedIn to differing degrees depending on the country.

Even if the numbers are not directly comparable internationally, LinkedIn assumes that AI trends are clearly represented on the platform since its members have an affinity to the digital domain. It is clear that even more needs to be done to promote women in the field of AI. The proportion of women among AI talents in Germany is relatively low compared with the other countries. Only 21% of the AI talents in Germany are women. In the USA, 28% of AI talents are women while the equivalent in the United Kingdom is 27%. In Germany, the majority of AI talents work in the ICT industry (31% compared with 12% of all LinkedIn users in Germany).

AI talents by industry (December 2020)

AI talents among LinkedIn users in 2020 according to industry classification

Note: LinkedIn data does not represent a randomised sample of the working population of a country. It is based on statements by LinkedIn users and as such does not necessarily represent all economically active persons in a country, even if a high proportion of those in many countries are LinkedIn users. The analyses presented here show the world from the perspective of LinkedIn, i.e. results are influenced by who uses LinkedIn and how. Depending on the country, this may be different and is influenced by various factors such as professional, socio-economic, regional or cultural background.

AI definition: AI talents are defined as any LinkedIn user who explicitly lists one of the AI skills identified by LinkedIn on his/her LinkedIn profile or who mentions one of the job titles defined by LinkedIn as an AI job. Examples of the skills identified by LinkedIn as AI skills include machine learning, natural language processing, data structures, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, Pandas (software) and OpenCV. Examples of AI jobs include machine learning engineer, artificial intelligence engineer and computer vision engineer. The complete list of AI skills and AI jobs is the property of LinkedIn and is not publicly available.

Methodology: The data shown here represents the level of data at LinkedIn at a specific time in December 2020. Based on LinkedIn’s economic graph, the LinkedIn users in the countries mentioned represent at least 40% of the working population, which vouches for the quality of the data. However, the quality of data for Germany is lower than the rest of the data due the lower penetration of LinkedIn among the working population. Although LinkedIn users do not specify their gender classification, this is derived from the first names of LinkedIn users. LinkedIn uses software that it programmed itself to perform this identification. The reporting industries in indicator 5.2. correspond to the classification used by LinkedIn. The assignment of LinkedIn industries to the Statistical Classification of Economic Activities in the European Community (NACE) can be helpful in classifying the LinkedIn sectors.

High demand for AI talents in Germany

The demand for more AI talents and AI skills is frequently a subject of discussion. It was also the subject of a study from 2019, which was commissioned by the Federal Ministry for Economic Affairs and Energy and conducted by the Leibniz Centre for European Economic Research. The study confirms the existence of strong demand for AI experts: 30% of companies that use AI in Germany had vacancies to fill in this field in 2019.

What picture emerges regarding the hiring rates of AI talents in other OECD countries? This indicator shows that the hiring rate of AI experts in Germany in 2020 was 1.58 times greater than in 2016. The greatest changes in AI hiring rates can be seen in Chile and in Turkey (possibly due to a lower starting rate). In Slovenia, the AI hiring rate in 2020 is lower than in 2016. 

Hiring rates of AI talents in selected OECD countries (December 2020)

The AI hiring rate index is indicated relative to the average AI hiring rate in 2016. Only OECD countries with 100,000 LinkedIn members or more are shown here.

Note: LinkedIn data does not represent a randomised sample of the gainfully employed population of a country. It is based on statements by LinkedIn users and as such does not necessarily represent all gainfully employed persons in a country, even if a high proportion of those gainfully employed in many countries are LinkedIn users. The analyses presented here show the world from the perspective of LinkedIn, i.e. results are influenced by who uses LinkedIn and how. Depending on the country, this may be different and is influenced by various factors such as professional, socio-economic, regional or cultural background. This indicator only shows OECD countries with 100,000 LinkedIn members or more.

AI definition: AI talent is defined here as any LinkedIn user who has explicitly listed one of the AI skills identified by LinkedIn on his/her LinkedIn profile or who has specified an AI job defined by LinkedIn in his/her curriculum vitae. Examples of the skills identified by LinkedIn as AI skills include machine learning, natural language processing, data structures, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, Pandas (software) and OpenCV. Examples that are classified by LinkedIn as AI jobs include machine learning engineer, artificial intelligence engineer, computer vision engineer. The complete list of AI skills and AI jobs is the property of LinkedIn and is not publicly available.

Methodology: LinkedIn calculates the hiring rate of AI talents for every country as follows: the number of AI talents who specified both a new job as well as a new employer on their LinkedIn profiles in a month in 2020, divided by the number of all LinkedIn users in a country. Further information can be found in the methodological note of the OECD AI Policy Observatory (OECD.AI).

Continuous increase in the hiring rate of AI talents in Germany

As the indicator “Hiring rates of AI talents in selected OECD countries” clearly shows, AI talents are in strong demand: their capabilities and skills are needed for artificial intelligence applications and there are too few of them on the labour market.

This scenario raises the question as to how demand for employees with AI skills has changed in recent years. Therefore, this indicator shows the rate at which AI talents are hired in Germany. The result: over the course of time, we can see that they have been hired in greater numbers since 2017, which clarifies the strong demand for AI talents.

Hiring rate of AI talents over time in Germany

Monthly 2017-2020

Note: LinkedIn data does not represent a randomised sample of the working population of a country. It is based on statements by LinkedIn users and as such does not necessarily represent all economically active persons in a country, even if a high proportion of those in many countries are LinkedIn users. The analyses presented here show the world from the perspective of LinkedIn, i.e. results are influenced by who uses LinkedIn and how. Depending on the country, this may be different and is influenced by various factors such as professional, socio-economic, regional or cultural background. In Germany, less than 40% of the working population are on LinkedIn.

AI definition: AI talent is defined here as any LinkedIn user who has explicitly listed one of the AI skills identified by LinkedIn on his/her LinkedIn profile or who has specified an AI job defined by LinkedIn in his/her curriculum vitae. Examples of the skills identified by LinkedIn as AI skills include machine learning, natural language processing, data structures, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, Pandas (software) and OpenCV. Examples that are classified by LinkedIn as AI jobs include machine learning engineer, artificial intelligence engineer, computer vision engineer. The complete list of AI skills and AI jobs is the property of LinkedIn and is not publicly available.

Methodology: LinkedIn calculates the hiring rate of AI talents for every country as follows: the number of AI talents who specified a new job and a new employer on their LinkedIn profiles in the same month in 2020, divided by the number of all LinkedIn users in a country. The AI hiring rates index is calculated as a moving average over twelve months in order to smooth the curve. Further information can be found in the methodological note of the OECD AI Policy Observatory (OECD.AI).

AI skills: Germany ranks in 2nd place internationally

How widespread are AI skills in Germany compared with other OECD countries? The indicator “Relative penetration of AI skills in OECD countries” – developed by LinkedIn and used by the OECD.AI Policy Observatory – shows that LinkedIn users in Germany specify AI skills 1.58 times as frequently as the OECD average. Compared with other OECD countries, Germany is therefore in second place behind the USA and ahead of Israel, Canada, South Korea, Japan, the United Kingdom and France. This shows that Germany is comparatively well positioned internationally with regard to AI skills, but has come catching up to do compared with the USA.

Relative penetration of AI skills in OECD countries

Comparison of relative penetration of AI skills among OECD countries with 100,000 LinkedIn members or more (average for 2015-2020)

Note: LinkedIn data does not represent a randomised sample of the working population of a country. LinkedIn data is based on statements by LinkedIn users and as such does not necessarily represent all economically active persons in a country, even if a high proportion of those in many countries are LinkedIn users. The analyses presented here show the world from the perspective of LinkedIn, which is influenced by how LinkedIn users make use of LinkedIn. This can depend greatly on the professional, socio-economic, regional or cultural background as well as on the availability and accessibility of the LinkedIn website.

AI definition: The definition of AI skills presented here covers a list of skills that were identified by LinkedIn as AI skills. This list includes machine learning, natural language processing, data structures, artificial intelligence, computer vision, image processing, deep learning, TensorFlow, Pandas (software) and OpenCV. The complete list of AI skills is the property of LinkedIn and is not publicly available. For more information about the LinkedIn Skill Group Classification, see also: World Bank Group, LinkedIn (2018), Data Insights: Jobs, Skills and Migration Trends Methodology & Validation Results, Appendix F.

Methodology: The indicator “Relative penetration of AI skills” is based on the Skills Genome Index developed by LinkedIn, which records the top 50 skills that are “both unique and representative” for a job title and then calculates the proportion of skills classified by LinkedIn as AI skills for this job title.

The penetration with AI skills in a country is defined as the average proportion that AI skills represent in the Skills Genomes of all job titles in this country. Since the list of job titles can vary from country to country, the indicator is not directly comparable across all countries. To enable an international comparison of the indicators, the relative penetration with AI skills is defined as the relationship between the penetration with AI skills in one country and the average penetration with AI skills in all countries. In this context, only job titles that occur in all countries are included. This indicator only shows countries with more than 100,000 LinkedIn users. It can be assumed that digital skills are relatively well represented on LinkedIn in the developed economies. Further information can be found in the methodological note of the OECD AI Policy Observatory (OECD.AI).

About LinkedIn

The professional networking service LinkedIn connects employees with employers and provides a platform for exchanging information, finding jobs, accessing further training opportunities and networking. All LinkedIn members have the option of listing their skills on their profiles. The member profiles can be matched with job advertisements that companies publish on the platform. There are currently around 740 million LinkedIn members worldwide and the platform advertises more than eleven million vacancies.

All LinkedIn members have the option of listing their skills on their profiles. The skills listed by members are self-evaluations although they can be confirmed by members in their personal network. There are currently more than 35,000 different, standardized skills that have been classified by LinkedIn. These are continuously updated with the help of algorithms that record the newly added skills of members. LinkedIn groups the skills specified by members into a taxonomy of 249 Skill Group Classifications, which are statistically evaluated in anonymised form for this data record.

The data on this career network does not cover the different areas of the German labour market to the same extent. Some sectors and regions are more strongly represented than others. In particular, the economic sectors “Industry”, “freelance, scientific, technical services” and “information and communication” are well represented. In contrast to official statistics, participation in the LinkedIn network and information shared by members is entirely voluntary.

Did you like this article? Share it with others: