Blog

Introduction

Artificial intelligence (AI) has made significant inroads into health care, demonstrating capabilities that complement and challenge traditional pharmacy practices. For instance, AI tools trained on clinical notes from electronic health records (EHRs) provided clinical predictions—including in-hospital mortality and 30-day readmission—with accuracy exceeding state-of-the-art risk scores.¹ AI interpretation of routine imaging (eg, chest x-rays) has the potential to provide comprehensive disease risk assessments, including risk of heart attack, stroke, and diabetes.² Electrocardiogram interpretation by an AI tool, flagging high-risk patients and notifying physicians, was found to reduce patient mortality.³ Additionally, patient assessments of responses to their questions found that those from AI chatbots were of higher quality than those from physicians.⁴

Turning to pharmacy, current and future AI applications include drug safety, pharmacy operations, precision medicine, drug reference navigation, clinical surveillance, and electronic clinical quality measures.^5,6 Looking forward, AI has the potential to use comprehensive patient-specific data such as EHRs, imaging, omics, and real-time monitoring data, in combination with medical domain expertise built on a foundation of medical literature to support caregivers in a variety of clinical tasks.⁷

As these AI-driven innovations continue to integrate into medical and pharmacy practice, critical questions emerge: What becomes of the pharmacist’s role? Where can the pharmacist fit into this new paradigm to provide value supporting safe and effective medication use? To see the path forward, it is helpful to first look backward.

Pharmacy’s Historical Adaptability

The pharmacy profession has a long history of evolving to meet health care needs. Over the decades, we have witnessed the emergence of specialized roles—drug information specialists, informatics pharmacists, and pharmacogenomics specialists—that did not exist previously but were created to address specific needs in the health care system. When a gap was identified between the health care team’s capabilities and the patient’s needs, pharmacists developed the drug information, information technology, and genomic expertise to fill those needs.

Illustrations of pharmacy’s adaptability are limited to not only the emergence of new roles but also the evolution of skills required for the role of any pharmacist. Before the emergence of electronic medical records (EMRs) and computerized physician order entry (CPOE), pharmacists were routinely tasked with interpreting handwritten prescriptions. Due to the influx of hurriedly scribbled prescriptions, pharmacists needed to accurately interpret such prescriptions to maintain efficient pharmacy operations. This skill, which was only tangentially related to medication expertise, became essential for pharmacists then. With EMRs and CPOE, this skill has all but vanished from the modern pharmacist’s arsenal. Computer skills are another example. While these had little value in pharmacy about 40 years ago, now they can dramatically impact the productivity of a pharmacist. Valuable skills for pharmacists will continue to evolve, and what present-day skills will become obsolete remains to be seen.

Adaptability has been a hallmark of the pharmacy profession, allowing pharmacists to remain integral to health care teams despite technological and systemic changes. As we stand on the brink of an AI alteration in health care, this adaptability will again be tested.

Emerging Gaps, Evolving Roles, and Preparing for an AI-Enhanced Future

As AI reshapes the health care landscape, new gaps will emerge between AI’s capabilities and patients’ needs. Given the plethora of possibilities, the challenges of regulatory approval, and the complexity of implementing new technology into health care delivery, it is nearly impossible to predict where AI will impact practice. If the dawn of AI in health care looks anything like the emergence of technology into any other industry, it will not fulfill its full potential in one fell swoop. Imperfect AI applications will emerge sporadically and improve iteratively over the years. This means that the gaps for pharmacists are likely unpredictable and unstable.

In the setting of this uncertainty, high-level skills that are broadly useful to a diverse set of scenarios will be most valuable. The following tasks are well-suited to leverage pharmacists’ clinical expertise while incorporating new technological competencies: AI education and implementation, loop oversight, human-on-the-loop oversight (quality assurance), and interdisciplinary collaboration (Table).^6,8 To thrive in this evolving landscape, the pharmacy profession must proactively prepare for an AI-enhanced future through education and continuous learning, hands-on experience, and advocacy and leadership.

Conclusion

The arrival of AI in health care presents challenges and opportunities for the pharmacy profession. Although some traditional roles may be transformed, pharmacists have the potential to adapt and evolve alongside these technological advancements.

By embracing change, acquiring new skills, and positioning themselves at the forefront of AI integration in health care, pharmacists can continue to fill crucial gaps in the health care system. The future of pharmacy in the AI era is about leveraging technology to enhance capabilities and improve patient outcomes.

The AI transformation in health care represents a significant shift in how we approach medication management and patient care. By actively engaging with these changes and helping to shape the integration of AI in health care, pharmacists can work toward maintaining a crucial role in the evolving health care ecosystem.

REFERENCES

1. Jiang LY, Liu XC, Nejatian NP, et al. Health system–scale language models are all-purpose prediction engines. Nature. 2023;619(7969):357-362. doi:10.1038/s41586-023-06160-y

2. Topol EJ. AI-enabled opportunistic medical scan interpretation. Lancet. 2024;403(10439):1842. doi:10.1016/S0140-6736(24)00924-3

3. Lin CS, Liu WT, Tsai DJ, et al. AI-enabled electrocardiography alert intervention and all-cause mortality: a pragmatic randomized clinical trial. Nat Med. 2024;30(5):1461-1470. doi:10.1038/s41591-024-02961-4

4. Ayers JW, Poliak A, Dredze M, et al. Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Intern Med. 2023;183(6):589-596. doi:10.1001/jamainternmed.2023.1838

5. Wong A, Wentz E, Palisano N, et al. Role of artificial intelligence in pharmacy practice: a narrative review. J Am Coll Clin Pharm. 2023;6(11):1237-1250. doi:10.1002/jac5.1856

6. Smoke S. Artificial intelligence in pharmacy: a guide for clinicians. Am J Health Syst Pharm. 2024;81(14):641-646. doi:10.1093/ajhp/zxae051

7. Moor M, Banerjee O, Abad ZSH, et al. Foundation models for generalist medical artificial intelligence. Nature. 2023;616(7956):259-265. doi:10.1038/s41586-023-05881-4

8. Nelson SD, Walsh CG, Olsen CA, et al. Demystifying artificial intelligence in pharmacy. Am J Health Syst Pharm. 2020;77(19):1556-1570. doi:10.1093/ajhp/zxaa218

Artificial intelligence sure has been taking a lot of flak lately.

Only 8.5% of the 48,000 people recently surveyed by accounting firm KPMG said that they “always” trust AI search results. Another report from Gartner found that more than half of consumers don’t trust AI searches, with most reporting “significant” mistakes.

A McKinsey study found that 80% of companies using generative AI have seen “no significant bottom-line impact”, with 42% of them literally abandoning their AI projects. An MIT study found that 95% of the AI pilot projects at the big companies they surveyed “failed”.

And now there’s workslop!

A new study published in the Harvard Business Review says that more than 40% of US-based full-time employees reported receiving AI-generated content that “masquerades as good work but lacks the substance to meaningfully advance a given task”. This “workslop” is “destroying productivity”, according to the study’s researchers.

Who is really to blame for workslop? Sure, blame big tech companies for yet again releasing untested and unproven products before they’re ready for prime time. Or the media and tech community who, for the past three years, have been writing pieces like Yahoo Japan wants all its 11,000 employees to use Gen AI to double their productivity by 2028 or AI will replace doctors, teachers, and make humans “unnecessary for most things”. All of this creates a lot of unnecessary hype and unfounded expectations.

But in the workplace, the buck always stops with the boss. The responsibility for AI’s “workslop” lies fully at the feet of the employer.

For more than 20 years, my company has implemented customer relationship and financial management applications at hundreds of small and mid-sized businesses across the country. We’ve worked with thousands of employees. We’ve had good projects and straight-out failures. As a technology consultant, we’ve made our share of mistakes. But the most common root cause of technology disappointments, failures and letdowns can always be found with the people who are buying and implementing the product.

So before throwing shade at software companies rolling out AI, I think it’s fair to ask employers a few questions.

For example, did you invest in training for your employees? Do your employees truly understand how to create the right prompts in order to get the best answers? Has your company standardized on an AI assistant or is it just a free-for-all mess of apps?

Do you have an AI policy that formalizes what AI can and cannot be used for and who can and cannot use it? Do you have a designated person in your company who is responsible for your AI-based applications? Has this person been trained and provided technical support to do this job? Are you working with a competent partner, consultant or developer to provide these kinds of services?

Most importantly, do you actually have a plan for using this technology effectively or are you just leaving it up to your employees to figure it all out? Do you have specific metrics for measuring AI’s effectiveness, or are you just relying on vague assumptions of “productivity”?

Unfortunately, many employers are duped by big tech into thinking that they just press a button and their software starts doing magical things that spew out money for their business. But, in order not to scare people away, these same tech companies don’t warn their customers of all the other things that need to happen – and money that needs to be spent – in order to maximize the use of their product. In most cases, the software is not the problem. It’s the lack of investment in the people using it.

AI can be a powerful tool if deployed the right way and with the right expectations. But in the end it’s just that: a tool. And new tools require thought, training, processes and investment. In the end, AI doesn’t produce “workslop”. Employers do.

Electricity generated from renewable sources has surpassed the amount of electricity generated from coal in Australia for the first time.

In September electricity from solar, wind hydro and biomass totalled 9.24 terawatt hours, compared with 8.8 terawatt hours from burning coal, according to data from the energy thinktank Ember.