AI’s top 10 jobs at risk: What happens when the technology predicts its own impact?

Artificial intelligence is becoming increasingly adept at performing tasks once considered the exclusive domain of humans. Yet an interesting question is emerging: if AI could assess its own impact on the labour market, which jobs would it identify as being most at risk?

A recent study by IT Asset Management Group (IT-AMG), sent to Digital Journal, attempted to answer exactly that. Researchers asked Google’s Gemini 3.5 Flash model a deceptively simple question: Which jobs do you believe AI is most likely to replace, and why?

The results provide a snapshot of how AI systems view the future of work. While the list should not be treated as a definitive forecast, it offers insight into the types of roles most vulnerable to automation—particularly those involving repetitive information processing, standardized decisions, and highly structured workflows.

The findings arrive amid continuing debate about whether AI will ultimately destroy jobs or create new opportunities. Speaking at the VivaTech conference in Paris in June 2026, Amazon founder Jeff Bezos argued that AI would not make people obsolete, predicting instead that technological advances would create labour shortages by unlocking entirely new industries and opportunities. Is Bezos correct or misinformed?

According to Gemini’s assessment, the occupation most vulnerable to automation is the data entry clerk. Data entry largely consists of transferring information between systems, processing forms, and validating records—tasks that modern AI systems and robotic process automation platforms already perform efficiently. As organizations seek productivity gains, many routine information-transfer roles are becoming increasingly automated.

Second on the list are telemarketers. Advances in conversational AI, voice synthesis, and automated call handling have dramatically improved the ability of systems to conduct scripted customer interactions. AI-powered voice agents can now answer questions, schedule appointments, and guide customers through basic transactions at a fraction of traditional operating costs.

Coming third are Tier 1 customer service representatives. Large language models can rapidly search internal knowledge bases, retrieve answers, and resolve common inquiries. Many organizations already deploy AI-powered chatbots as first-line customer support, reserving human agents for more complex cases.

The fourth-ranked occupation is bookkeeping. Routine reconciliation, expense categorization, invoice processing, and transaction monitoring increasingly fall within the capabilities of AI-enhanced accounting software. Human expertise remains necessary for audits, interpretation, and financial strategy, but much of the transactional work is becoming automated.

Rounding out the top five are proofreaders. Grammar-checking systems and AI writing assistants have become remarkably capable at identifying spelling errors, punctuation issues, stylistic inconsistencies, and readability concerns, reducing the demand for basic proofreading services.

Perhaps more interesting are the roles appearing lower in the rankings. Paralegals, especially those engaged in document review, were ranked sixth. Legal teams frequently manage thousands of pages of contracts, case files, and regulatory documents. AI systems can rapidly search, classify, summarize, and compare these records, potentially reducing the size of document-review teams.

In seventh place are commodity content writers. The distinction is important. AI is not necessarily replacing investigative journalists, technical experts, or original storytellers. Rather, the technology excels at generating high-volume content such as product descriptions, basic blog posts, and routine marketing copy.

Market research analysts also appear on the list. Data collection, sentiment analysis, competitor monitoring, and report generation are increasingly automated. However, strategic interpretation and decision-making still require human judgment.

Ninth are retail cashiers, reflecting the continued expansion of self-checkout systems and cashier-less retail technologies. Although widespread replacement requires substantial infrastructure investments, the technological pathway is clearly established.

The final position is occupied by commercial translators. Advances in multilingual AI models have produced dramatic improvements in machine translation quality. While human translators remain critical for nuanced, legal, literary, and culturally sensitive material, routine business translations are increasingly being automated.

What connects these jobs?

A common theme runs throughout the top ten. Most involve tasks that are repetitive, rules-based, and highly structured. Furthermore, these roles are dependent on processing information, and they tend to be measured by speed and consistency. These characteristics make them particularly suitable for automation.

Notably absent are occupations requiring emotional intelligence, complex physical dexterity, ethical judgment, relationship building, leadership, creativity, or deep contextual understanding. The technology continues to struggle whenever ambiguity, uncertainty, or human trust become central to performance. This distinction helps explain why AI can draft a customer service response but cannot easily replace a skilled therapist, nurse, scientist, negotiator, or senior business leader.

Will jobs disappear or evolve?

History suggests that technological disruption rarely results in simple replacement. Previous waves of automation eliminated certain tasks while creating entirely new occupations. The introduction of personal computers reduced demand for typists but created entire industries in software development, cybersecurity, digital marketing, data science, and IT support.

Many economists believe AI will follow a similar pattern. While routine aspects of jobs may disappear, humans may increasingly focus on oversight, interpretation, quality assurance, relationship management, and strategic decision-making.

Indeed, AI itself acknowledges this limitation in several of the occupations it identified. Bookkeepers may become financial analysts. Paralegals may shift toward case strategy. Customer service representatives may handle only complex or emotionally sensitive interactions.

The Canadian perspective

For Canada, these developments carry significant implications. The country has invested heavily in artificial intelligence research, hosting globally recognized AI centres in Toronto, Montreal, Edmonton, and Vancouver. As adoption accelerates, Canadian organizations are likely to face both productivity opportunities and workforce disruption. Sectors heavily reliant on administrative processing, such as finance, insurance, legal services, customer support, and government administration, may experience the fastest transformation. At the same time, demand could increase for AI trainers, prompt engineers, data governance specialists, cybersecurity professionals, digital ethicists, and technology auditors.

Bezos offers an unduly optimistic outcome. The extent of disruption will depend on how effectively societies manage the transition.  The lesson from Gemini’s list is not that jobs will vanish overnight. Rather, it highlights which tasks are most susceptible to automation and where workers may need to develop new skills.