The rise and impact of AI in Biotechnology and Healthcare

AI is here to stay, transforming all industries at varying speeds. It is rapidly evolving from a tool to a driver of efficiency, innovation, and decision-making. While its impact is acknowledged, opinions differ on workforce effects: some see AI as augmenting human work, and even as teammates, but others fear complete replacement by systems capable of all human functions. There is a growing call for ethical considerations, governance, and transparency to address bias, data privacy, and societal inequalities - including gender inequality.

The global market for AI in biotechnology is expected to surge from an estimated US$3.8 billion in 2024 to about US$11.4 billion by 2030.

This growth is being driven by AI-powered advances across drug discovery, diagnostics, personalised medicine, and biomanufacturing AI transformation is evident across all business units, including its core R&D. It profoundly impacts all functions, allowing the entry of disruptive competitors and introducing new complexities into market dynamics and interorganisational relationships.

Understanding AI-agents

AI is transforming the world and is an endlessly evolving field. It has shifted from narrow-task systems to autonomous architectures. Among these advancements, AI-agents represent a fundamental change in how AI systems operate. They represent a specialised field within the broader AI umbrella, going beyond the task-specific instructions required by earlier AI systems. AI-agents combine perception, decision-making, and action, representing a significant leap in AI autonomy.

AI-agents leverage a combination of Machine Learning (ML) techniques [4], including:

• Natural Language Processing (NLP): Enables AI-agents to summarise text, translate languages, and interpret user intent, making AI-agents' interactions more human-like.
• Reinforcement Learning (RL): Agents trained with RL learn by interacting with environments, receiving rewards or penalties based on their actions.
• Deep Learning (DL): AI-agents are based on neural network architectures and transformer-based models that enable advanced pattern recognition, contextual memory, and reasoning, especially through generative models (GenAI), which further facilitate content creation (e.g., text, images).
• Large Language Models (LLMs): GPT-based systems (core to LLM-based agents – Agentic AI) enable advanced comprehension and problem-solving.

They can operate as a single-agent or multi-agent system (where specialised AI-agents collaborate with each other using communication and negotiation protocols and distribute decision-making to exceed the capabilities of individual agents). AI-agents can be categorised based on various factors, including behaviour and ability to handle complexity. Regardless of the AI-agent type, they all share three core components: architecture (the base from which the AI-agent operates), function (that describes how data is translated into actions), and program (that implements the AI-agent's function).

The strengths of this technology include:

• Autonomy, based on built-in decision-making frameworks to respond to environmental cues, enabling performance with minimal supervision.
• Analytical intelligence, based on advanced algorithms, that break down problems, analyse data, and make logical decisions/deductions.
• Tool use/calling capabilities to connect with APIs, databases, computational tools, and physical devices.
• Adaptability, as they learn from new data and feedback, and adjust their behaviour accordingly in real-time.
• Efficiency and goal orientation, as they manage large volumes of data (even in environments with ambiguity and incomplete data), streamline workflows and automate decisions.
• Contextual awareness, as they retain memory across interactions, leading to better interpretation and more relevant responses.
• Scalability and interoperability, allowing them to collaborate with each other and integrate with enterprise systems.

However, they also have weaknesses/challenges such as:

• Reasoning limitations, with AI-agents vulnerable to fallacies and with difficulty with multi-step deductions.
• Context management, especially over extended periods and across multiple sessions, as AI-agents often lack robust long-term memory.
• Tool integration issues, as existing systems may lack appropriate APIs or data format, leading to more complex AI-agents integration. In addition, AI-agents may select inappropriate tools for specific tasks, misinterpret their outputs, and struggle with interface changes, limiting their autonomy.
• High resource requirements, demanding high computational capabilities and costs for training and deployment.
• Robustness issues, including hallucination (the generation of false or misleading information), inconsistency (the production of contradictory responses), and unpredictable performance under stress. Connected to this is a growing concern with agent hijacking.
• Limited explainability and transparency, as many AI-agents operate as "black-boxes", making their decision-making processes difficult to understand.
• Ethical risks, with growing concerns around fairness, privacy, and responsibility/accountability.

Currently, AI-agents are a form of Artificial Narrow Intelligence (ANI), but their growth signals a pathway toward Artificial General Intelligence (AGI).

In the workforce, AI-agents are expected to evolve from supportive assistants into active collaborators - transforming how work is executed, redefining skill requirements, and reshaping decision-making and value creation. This will establish AI-augmented environments across industries that will require robust governance and ethical balance throughout the value chain.

The Three Key Shifts in Biotech

AI is poised to reshape the biotech industry through the following shifts:

1.  Radical acceleration of early discovery: This involves the rapid acceleration of R&D activities. AI transforms research, with machine learning enabling faster data processing, analysis, and literature reviews. Predictive analytics identifies promising research avenues while minimising time spent on unproductive studies. Platforms like AlphaFold 3 predict complex 3D protein structures in hours - previously requiring months or years - accelerating drug development. "Self-driving labs" establish closed-loop experimental cycles where AI designs experiments [35], robotic systems execute them, and algorithms analyse results to design subsequent iterations [34]. AI lab assistants enable scientists to use natural language commands for complex multi-step protocols across connected instruments [15, 16, 37]. This shift transitions researchers from lab work to engaging in strategic thinking, while requiring them to develop stronger digital and AI fluency [34].

2.  The rise of competitors with AI-based business models and talent recruitment disruption: The transformation of R&D marks only the beginning - AI is now influencing the entire biotech value chain. It optimises production (AI-powered Manufacturing Execution Systems reduce labour costs and defects, predictive analytics identifies bottlenecks and enhances processes, and predictive maintenance monitors equipment failures, minimising costly downtime.

 AI is being used to improve trial design, patient selection and operational efficiency, with early evidence of higher success rates in selected phases and use cases. This includes AI-platforms that reduce manual patient screening, wearable AI devices that enable real-time data monitoring, automated analysis of health records for immediate adverse event detection. AI also makes supply chains more sustainable and robust by analysing demand, production schedules, and geopolitical factors to adjust inventory proactively. Cheaper, faster, and more reliable AI-driven value chains reduce entry barriers in this market. New entrants using AI throughout their value chain operate with smaller teams and faster R&D cycles, increasing competitive pressure and further transforming the industry and workforce. These companies adopt alternative talent acquisition strategies by targeting hybrid professionals with biological and computational expertise. They use AI-driven tools to find promising candidates from unconventional backgrounds. Their compensation models differ from traditional salary structures, often including significant equity and stock options, which, along with flat organisational structures, attract professionals eager for impact.

3.  High demand for AI governance and data security: The evolving market landscape shaped by AI-integrated companies requires comprehensive governance frameworks. The use of AI to process large volumes of sensitive data (e.g., digital twins - virtual AI-powered replicas - employed to optimise patient recruitment and predict clinical outcomes by analysing genetic data, medical records, and proprietary research) creates the need for robust security infrastructures. This has led to an increase in regulations, such as the Australian government releasing AI Ethics Principles and the EU AI Act establishing new global standards for transparency and accountability. Regulators and agencies in the US have tightened expectations and scrutiny around AI and data security, particularly in pharma. The need for governance is augmented by biased datasets, which cause disparities in disease predictions and treatment, particularly impacting underrepresented groups and raising ethical and social concerns that need specialised oversight. This governance imperative generates new job categories, such as "algorithmists" and AI & Ethics Governance specialists, with AI-skilled workers earning a significant wage premium. 

AI-agents in healthcare

From an industry perspective, AI-agents primarily impact organisations that are highly regulated, have large volumes of data, complex decision-making needs, and real-time operational dependencies. In healthcare, AI-agents are changing organisations in this sector that rely on precision, timeliness, and data integration, including hospitals, clinics, and diagnostic laboratories. AI-agents streamline workflows, enhance patient care, and optimise operational efficiency.

One use case is virtual health assistants, which offer 24/7 patient support, including medication reminders, appointment scheduling, and symptom checking. This reduces administrative burden and improves patient compliance.

In chronic disease management, wearable-data monitoring agents analyse patient data to detect health deterioration and coordinate interventions. This reduces hospital readmissions and enhances long-term health outcomes. A study with type 2 diabetes patients using an AI-agent showed significant benefits: medication adherence rose by 27%, dietary compliance increased by 32%, and overall glycaemic control improved, with an average HbA1c reduction of 0.8%, compared to standard care methods.

In diagnostics, imaging-diagnostics agents are used to analyse medical images and detect conditions with precision often exceeding that of human clinicians. This has enabled faster, evidence-based clinical decisions. A major academic medical centre implemented an agent-based clinical decision support system that improved diagnostic accuracy for complex cases by 23% and reduced time to diagnosis by 37%.

Overall, the ability of AI-agents to integrate multimodal data (e.g., clinical notes, sensor readings, imaging) makes them powerful tools in supporting healthcare operations, enabling scalable and precise care delivery.

Embracing the future with responsibility

AI is here to stay, fundamentally transforming industries globally, with biotech and healthcare at the forefront of this revolution. The radical acceleration of R&D, the emergence of new AI-driven competitors, and the critical demand for robust governance and data security are reshaping the industry landscape. AI-agents, powered by advanced ML techniques, offer unprecedented capabilities in autonomy and efficiency, yet they also pose challenges related to reliability, transparency, and ethical considerations.

From improving patient care in healthcare to optimising financial fraud detection, AI agents are proving to be powerful tools. However, their widespread influence means that organisations must proactively consider the impacts on stakeholders such as employees and customers, balancing efficiency gains with ethical responsibilities and human considerations. The future of AI in the world depends on navigating these complex changes thoughtfully and strategically to maximise AI's full potential while minimising its risks for a more ethical and efficient future.

Telma Mantas is Director, IP and Licensing at Cartherics, a biotechnology company developing off-the-shelf immune cell therapies focusing on high-impact women's diseases, with lead programs in ovarian cancer and endometriosis.