The year 2023 seemed to herald the widespread adoption of AI in the daily lives of employees, so what about a year later?
This integration appears to be much less straightforward, with the AI Act passed by the European Parliament on May 13 now providing a framework for the use of this technology without curbing the thirst for innovation.
One indicator to watch is the slowdown in traffic to the ChatGPT app. It has stabilized at around 180 million monthly users, with growth of just 13% between March 2023 and March 2024.
Individual employee ownership contrasts sharply with strategic integration at corporate level. Despite the fact that AI is among the top five priorities for large companies, only just over 10% have undertaken a meaningful impact study on their businesses.
How can we explain this moderation in the uptake of these technologies? What uses are they really being put to today, and what projections can we make for tomorrow?
The different forms of AI adoption
The adoption of AI is not monolithic. Our many discussions with HR managers and business leaders show that its spread varies from use for productivity gains to its integration into the company's own offering.
Let's take a closer look at these different forms of adoption, their respective maturities and their obstacles. We also consider it essential to differentiate these elements according to the type of AI in question.
First use: improving individual employee productivity.
Artificial Intelligence for personal use : using AI systems as intelligent personal assistants to optimize daily tasks and time management.
- Maturity: High. These technologies are increasingly in demand in technology-intensive sectors and by jobs requiring efficient time management.
- Barriers: The cost of technology and the varying quality of the tools available can represent obstacles, particularly for SMEs and less technology-oriented sectors.
Second use: optimizing business processes.
Automated management systems: Deploying AI solutions in supply chain management, robotic process automation (RPA) and customer relationship management (CRM) systems that transform internal operations and improve overall efficiency.
- Maturity : Medium. Large companies, particularly in the manufacturing, retail and financial services sectors, are widely integrating these systems to optimize internal operations.
- Barriers: The high initial investment and the need for ongoing maintenance and integration with existing IT systems may limit their selection by SMEs.
Third use: making technology available to end customers.
AI integrated into product/service offerings: these new services better meet customer needs, such as personalized recommendations or intelligent user interfaces.
- Maturity: Rapidly growing. Particularly present in consumer-oriented industries such as retail, entertainment and telecommunications.
- Barriers: Requires in-depth understanding of customer data and strong data analysis capabilities, which may limit use to companies with insufficiently robust data infrastructures.
Two illustrations of advanced integration in products and services are worth highlighting here:
1st illustration: copiloting with Neobrain
Neobrain has sealed a partnership with Microsoft to integrate "co-piloting" into its solution, more specifically in the workforce planning dimension. Users can ask their assistant questions at any time to obtain detailed answers and access data, thus avoiding long and tedious searches in their application.
Video illustration:
2nd illustration: Sage Productivity Assistant
British software publisher Sage. Its People Director for Southern Europe, Tiphaine Brisou-Debeze, told me in an interview about the launch in early March of their generative AI-powered productivity assistant. On the other hand, when I ask her if her employees ask her for access to Chat GPT, the answer is clear: "none".
But how can we be sure that employees aren't practicing a new form of shadow IT through the use of "shadow AI"?
What is shadow AI?
Shadow AI refers to the use of artificial intelligence applications and services by employees without the approval or control of corporate IT departments. This practice can lead to security and non-compliance risks, as it bypasses the company's official data management and security policies.
Many large companies such as Samsung, Apple, Deutsche Bank, Verizon and Amazon have banned ChatGPT to prevent data leakage. As a result, these companies need to find solutions to offer the same level of service in a secure way.
How can we tell the different AIs apart?
Artificial intelligence covers a wide range of modalities, which vary not only historically but also in terms of data processing protocols and ease of access.
Let's take a closer look:
- Machine learning is crucial, but often limited to large companies with extensive databases and robust IT systems, as it requires a significant volume of data to be effective.
- Natural language processing (NLP) is more widely used, particularly in human resources for screening job applications and analyzing behavior and communications. Its ability to handle natural language makes it accessible and useful in many professional contexts.
- Wide language models (LLMs), such as ChatGPT, find widespread application in marketing, creative, legal and sales roles thanks to their ability to generate textual content autonomously. However, their use is less obvious in finance or manufacturing. This is due to the specific requirements for excessively reliable data and precise calculations, a level of confidence in the data that is not always achieved.
Key AI adoption statistics
To assess the real development of artificial intelligence in the workplace, we need to take stock and compare different statistics.
Modelling is still subject to many uncertainties, as evidenced by the differing points of view of major studies. For example, at the end of 2023, McKinsey stated that "AI could replace around 300 million full-time jobs in the future", while Forrester spoke of 2.4 million jobs directly impacted by 2030, and 11 million other jobs influenced.
Current adoption statistics
Company penetration (source: United States Census Bureau)
- 34% of companies have implemented AI technologies
- 42% are currently exploring the AI options available.
- 35% of organizations are actively training and reskilling their teams to make effective use of new AI and automation tools.
Penetration by country (source IBM):
- India is one of the countries with the highest use of Generative AI, at 57%.
- Canada stands at 48%.
- The United States at 25%.
- France at 31%.
Impact on occupations:
Our study "How does AI impact my workforce", carried out in January 2024, reveals 4 levels of impact on professions:
- High impact" category (32%) :
- The jobs undergoing a notable transformation due to AI, with 40 to 66.6% of their business affected.
- Medium Impact" category (36%) :
- jobs where the impact of AI is moderate, with certain activities able to benefit from AI (20 to 40% of impacted activities).
- Limited impact" category (18%) :
- The jobs who are experiencing a marginal transformation of their tasks due to AI (10-20% of activities impacted).
- No impact" category (14%) :
- The so-called "protected" jobs , where the very nature of the tasks remains unchanged by the AI.
Identify the degree of automation of 100 professions in our web-app:
Statistics on future adoption projections
Investment forecasts :
- 39% of large companies (+10,000 employees) plan to step up their investment in AI for their employees (Adobe)
The activities that consume the most AI :
- The top 5 functions and activities that will benefit most from AI :some text
- Customer service
- Cybersecurity, fraud detection
- Customer relationship management
- Inventory management
- Content creation
- Recruitment ranks 10th according to Forbes Advisor.
Expected benefits :
- Competitiveness and productivity are the 2 driving forces behind its deployment, with 87% and 64% respectively (Forbes).
A growing appetite for training:
- 67% of employees would like their company to support them in acquiring skills prompting AI.
Gaps to close skills :
- 70% of managers believe that their teams are not sufficiently qualified to optimize the use of AI. 61% of these same employees say they do not have the necessary skills to use generative AI effectively and safely (Salesforce GenAI snapshot research studies).
Conclusion
The past year has highlighted both advances and persistent challenges, reflecting a complex but inevitable transition to AI-enhanced work environments. It's important to underline a certain kind of aspiration to capitalize on these technologies for a portion of white-collar workers, as the experience of interacting with the tool definitely revolutionizes past software in the image of what co-pilot offers, for example. This is good news for HR teams, for whom getting employees to use their HR software is not always a sine cure.
This gradual preference will undoubtedly be more difficult for an older population, less inclined to see the value of training in this area. Which raises the question of the ongoing redeployment of staff.
Our next article on the subject of AI investigates the obstacles and examples of companies that have initiated robust rethinking projects, to propose a 5-step methodology for initiating this inescapable change.
