Not in the mood to read? Here's the Promptcast
(Recap generated with NotebookLM, with minor edits)
Dr. Sami Bahri's analysis of the current AI landscape points to a decisive turning point: the shift from mere hype to serious business use. While media attention is often hijacked by spectacular “proofs of concept”, the real economic value lies in automating knowledge worker processes.
Key takeaways:
- Technological maturity: On the Gartner Hype Cycle, we are moving away from the peak of inflated expectations towards stable business solutions. Even so, accuracy in critical fields such as law (59 %) or medicine (~45–49 %) remains problematic. With around 80 % accuracy, coding is the current “killer app”.
- The Gen Z dilemma: A “workforce shifting” is under way. Seniors with domain expertise, combined with AI agents, are more productive and cheaper than juniors, which makes entering the job market massively harder for the next generation.
- Localisation and data protection: The trend is moving away from the cloud towards private large language models (LLMs) to preserve the integrity of company data (“AI knows the Internet, it doesn't know your enterprise”).
- Creative Destruction: Between 30 % and 50 % of the EU labour market will be hit by massive restructuring or job losses in the medium term (5–10 years). This calls for urgent adjustments in education policy and in business.
1. Status quo: between signal and noise
The current phase of AI development is marked by a gap between public perception and actual business value.
- Hype cycle and market situation: According to Dr. Sami Bahri, we are currently in a phase where the “noise” often drowns out the actual “signal”. The massive investments in the US (the “Trillion Dollar Bet”) are viewed critically as a potentially unhealthy bubble that combines elements of the 2001 tech crash and the 2008 debt crisis.
- Business-ready vs. POC: Many existing solutions are merely impressive demos, not yet market-ready tools. AI won't replace existing IT tools; it will join them in the toolbox as a “new high-performance drill”.
- Attention economy: Big vendors deliberately try to capture users' attention in order to harvest data for training their models.
2. Technology trends and benchmarks
There is no such thing as “the one AI”. The market is increasingly splitting into specialised models.
Performance by sector
The accuracy of AI models varies widely depending on how structured the underlying data is:
Field | Hit rate / score | Note |
Coding | 77% – 80%+ | Killer app; structured languages are easy for AI. |
Legal | approx. 59% | High risk of error when used uncritically. |
Finance & Accounting | approx. 50% | Often unreliable with complex financial data. |
Healthcare | 45% – 49% | Dangerous for diagnoses without human oversight. |
Key technological developments
- Local LLMs: To prevent exfiltration of company data, local models are gaining ground. They can run directly on knowledge workers' hardware (recommendation: workstations with ≥ 64 GB VRAM) or in a private company cloud.
- Second Brain: AI-powered knowledge stores enable automatic tagging, tables of contents and filtering of information from large document collections. An ideal tool for knowledge workers.
- Agentic workflows (Claws): Orchestrating specialised AIs via messenger interfaces (e.g. WhatsApp, Discord, etc.) that work through chains of tasks autonomously and have a “memory”.
- Tokenomics & token-maxing: A worrying trend in which AIs generate needlessly long answers (“token-maxing”) to artificially inflate the consumption of compute units (tokens) and thus the cost to the user.
3. Economic transformation: creative destruction
Following Schumpeter's principle of “Creative Destruction”, AI integration is driving a radical change in the world of work.
Business Process Optimization (BPO)
The challenge lies in automating knowledge worker processes. Logistics processes are deterministic; knowledge workers, by contrast, often work probabilistically (weighing options).
- AI massively boosts productivity in knowledge work with unstructured data.
- The focus shifts from completing simple tasks to the quality of decision-making.
The Gen Z dilemma and workforce shifting
Companies are facing a strategic realignment of their workforce:
- Seniors: Have the expertise needed to validate AI output. Paired with AI agents, they are more efficient than junior teams.
- Juniors (Gen Z): For companies, the cost of training new entrants becomes too high compared with using AI tools. The result: fewer graduate and entry-level hires.
- Consequence: A massive cut in junior and assistant positions, alongside a reorganisation of the remaining workforce.
4. The role of education and what society needs to do
Technological change is acting as an “accelerant” for social and structural problems.
- Role shifting in education: Classic theory teaching at universities is losing relevance, because AI makes theory easily available. What's needed is a “seniorisation of juniors”. Experienced practitioners must step in as the new teachers, passing on practical know-how in working with AI.
- Political implications: 30–50 % of the EU job market will be affected going forward. Political systems often only react once problems become visible. Dr. Sami Bahri stresses the need to tackle the “Social Innovator’s Dilemma” and to create new frameworks for the labour market fallout.
- Infrastructure conflicts: The enormous resource demands of data centres are already causing local conflicts over power and water (especially in the US). Forecasts suggest AI data centres could claim a significant share of national electricity consumption.
5. Strategic recommendations
To meet the transformation head-on, the following measures are proposed:
- Process focus: Document and define internal knowledge worker processes so they become accessible to AI automation (BPO).
- Senior enablement: Targeted training programmes for experienced staff (seniors) to combine their domain expertise with AI skills.
- Local data storage: Prioritise private AI infrastructure for sensitive company data to protect intellectual property.
- Early adaptation: Recognise that “Creative Destruction” is already under way and calls for a proactive redesign of role profiles within the company.
