AI Research

How generative AI can make accountants more productive

Published

1 month ago

August 5, 2025

zach church

What you’ll learn:

AI amplifies accounting experience: In a study of accountants, the most experienced used AI strategically to boost performance gains.
Productivity and reporting quality both improve with AI: Working with AI-enabled software helped accountants support more clients and finalize monthly statements sooner.
Expertise still matters: Human judgement is needed to evaluate AI’s work is and avoid errors.

+++

A new study confirms what some accountants may have already known: Generative artificial intelligence software is making them more productive and, in many cases, improving the quality of their reporting.

MIT Sloan assistant professor along with Stanford University’s Jung Ho Choi, partnered with a company that makes AI-based accounting software so they could analyze hundreds of thousands of transaction entries from 79 small and midsize firms. They also surveyed 277 accountants, about 10% of whom were using AI in their daily routines; another 10% had not experimented with AI at all.

A majority of the accountants surveyed felt that AI provided efficiency benefits and has the “potential to reduce repetitive work and improve analysis,” Xie and Choi write. And a majority felt that it would increase their job satisfaction, as it pertains to productivity, work-life balance, and their careers. While the research occurred early in the study of generative AI and accounting, its findings suggest practical implications for accountants, managers, and accounting firms.

Clear productivity and quality gains

The software in the study automates routine work and can help accountants make decisions. For example, it can classify transactions, summarize contracts, and detect anomalies in bookkeeping. Accountants using the software saw greater productivity on average, including:

An increase in weekly client support.
A “reallocation of approximately 8.5% of accountant time from routine data entry toward high-value tasks such as a business communication and quality assurance,” the researchers write.

Those same accountants also saw improved financial reporting quality, seen as:

A 12% increase in general ledger granularity (a measure of reporting detail).
A 7.5 day reduction in monthly close time.

“Essentially, firms embracing AI are able to finalize their monthly financial statements almost within two weeks after month-end, whereas others take over a week longer,” Xie and Choi write.

The researchers found that many accountants were using the tool to handle routine work, freeing up time for analysis and work with clients. They also found that “more experienced accountants tend to leverage the AI system more strategically and reap larger performance gains from it.” It’s possible that the more experienced accountants are better at interpreting the confidence scores the AI software applies to its own recommendations and thus are more likely than their less-experienced peers to intervene when the scores are low, they write.

Concerns about AI accuracy

Many accountants surveyed said they had concerns about using the AI software, including 62% who were worried about errors and accuracy in AI-generated reporting. Accountants were also concerned about data security issues and job displacement. One area of concern for the researchers: “When AI suggests diverging categories for uncertain transactions, accountants tend to still follow AI’s suggestions,” introducing the possibility of errors attributable to AI.

What does this mean for accounting firms?

For managers: When integrating AI-assisted accounting systems, recognize that the technology works best when it augments your existing experts. Accounting isn’t just following a set of rules, Choi said. As powerful as AI is, it isn’t always able to consider all of the context around information. For example, when AI confidence scores are low, judgment from experienced accountants is needed.

For accountants: Consider using AI to automate much of the grind and boost job satisfaction. More experienced accountants may see even greater gains.

“There’s that really famous meme, ‘What I want AI to do is do my laundry so that I can write poetry, not write poetry so I can do laundry,’” Xie said. “In accounting, there’s laundry and there’s poetry.” Make the laundry — inputting and processing data, for example — more efficient and you free up time and space for the poetry, she said. That means more time for client interactions, financial forecasting, higher-level thinking, and the like.

“I think intrinsically, as human beings, we want to do creative judgment-based work,” Xie said.

For the profession: Getting AI and accountants to work together well will require AI literacy training, and clear oversight standards are needed to scale the net gains of AI. Xie and Choi’s research examines AI in accounting as it is employed now, and it poses questions for future work: How should organizations prepare for a generation of accountants who have never done accounting work without AI? What tasks or roles will still exist? And how should managers think about organizational structure for the accounting function?

Read the research

The full research paper, “Human and AI in Accounting: Early Evidence from the Field,” discusses AI adoption patterns, task reallocation, and how experienced accountants use AI confidence scores.

This article is based on work by Chloe Xie, an assistant professor of accounting at MIT Sloan. Her research focuses on capital market imperfections — such as limits to arbitrage, deviations from von Neumann-Morgenstern preferences, and criminal behavior — and how these frictions shape the information environment. Her research also considers how these frictions affect disclosure decisions, asset pricing, investor decision-making, and non-financial market outcomes.

Source link

AI Research

How AI Is Transforming Disease Research and Drug Discovery

Published

48 minutes ago

September 16, 2025

Julian Horsey

What if the cure for cancer, Alzheimer’s, or genetic disorders was hidden in plain sight, buried within mountains of data too vast for any human to process? In an era where scientific progress is often limited by the sheer volume of information, artificial intelligence is stepping in as a fantastic option. Enter Sam Rodriques, a scientist at the forefront of this revolution, whose work explores how AI can transform disease research. In this thought-provoking exchange with Freethink, Rodriques sheds light on the innovative tools reshaping medicine, from multi-agent AI systems to new applications in drug discovery. Could AI not only accelerate research but also redefine how we approach the most complex biological puzzles?

Below Freethink uncover how AI is addressing the limitations of human cognition, automating labor-intensive processes, and fostering collaboration across disciplines. Rodriques offers a rare glimpse into the development of specialized AI agents like Crow and Phoenix, each designed to tackle specific stages of research, from synthesizing literature to planning experiments. But this isn’t just about technology; it’s about the human ingenuity guiding these tools and the ethical questions they raise. Whether you’re curious about the future of medicine or the role of AI in shaping it, this dialogue promises to challenge assumptions and inspire new ways of thinking about scientific discovery. What happens when machines and minds work together to unlock the secrets of life itself?

AI Transforming Scientific Research

TL;DR Key Takeaways :

AI is transforming scientific research by automating complex tasks, generating data-driven hypotheses, and integrating knowledge across disciplines, particularly in biology and medicine.
Multi-agent AI systems, such as Crow, Falcon, Finch, Owl, and Phoenix, collaborate to streamline workflows, enhance precision, and accelerate research processes.
AI-driven research emphasizes transparency and traceability, making sure findings are grounded in empirical data and fostering trust within the scientific community.
Real-world applications, such as AI-generated hypotheses for treating diseases like age-related macular degeneration, demonstrate AI’s potential to bridge theoretical insights and practical outcomes.
While AI offers fantastic potential, it requires human oversight to address challenges like ethical considerations, data limitations, and context-dependent scenarios, making sure responsible and effective use in research.

The Growing Need for AI in Science

Modern research generates an overwhelming volume of data, making it increasingly challenging for researchers to synthesize information and extract actionable insights. AI offers a powerful solution by automating repetitive tasks such as literature reviews, data analysis, and hypothesis generation. These tools are not designed to replace human expertise but to complement it, allowing researchers to explore scientific questions more efficiently and comprehensively.

For example, AI can integrate findings from diverse disciplines to propose innovative approaches to treating diseases or understanding complex biological systems. This capability is particularly valuable in addressing challenges such as drug discovery, where identifying potential compounds and predicting their effects require analyzing massive datasets. Similarly, AI is instrumental in unraveling the intricacies of genetic disorders, where patterns in genomic data may hold the key to new treatments.

Multi-Agent AI Systems: A Collaborative Approach

One of the most promising advancements in AI-driven research is the development of multi-agent systems. These platforms consist of specialized AI agents, each designed to excel in a specific task, working together to automate complex workflows. By delegating tasks among these agents, researchers can achieve faster and more accurate results. Key examples of these agents include:

Crow: A general-purpose agent that synthesizes literature-informed science, providing a broad foundation for research.
Falcon: Specializes in conducting deep literature searches and performing meta-analyses to uncover hidden connections.
Finch: Focused on data analysis and hypothesis testing, making sure that conclusions are grounded in robust evidence.
Owl: Conducts precedent searches to evaluate the novelty and feasibility of new ideas.
Phoenix: Excels in experimental planning, particularly in chemistry, by designing experiments that maximize efficiency and accuracy.

These agents operate collaboratively, with each contributing its expertise to different stages of the research process. For instance, one agent might analyze existing literature to identify gaps in knowledge, while another designs experiments to address those gaps. This division of labor not only accelerates the research process but also enhances the precision and reliability of the outcomes.

Sam Rodriques on AI’s Potential to Cure Cancer and Alzheimer’s

Gain further expertise in Artificial Intelligence in Science by checking out these recommendations.

Transparency and Traceability in AI-Driven Research

In scientific research, transparency and traceability are critical for making sure trust and reliability. AI systems address these requirements by providing detailed reasoning, citations, and traceable workflows. As a researcher, you can review the evidence and logic behind AI-generated conclusions, making sure that findings are grounded in empirical data and aligned with established scientific principles.

This level of transparency reduces the risk of errors and enhances confidence in AI-driven discoveries. It also allows researchers to scrutinize and validate AI outputs, maintaining the rigor of the scientific process even as automation takes on a larger role. By allowing traceability, AI systems ensure that every step of the research process can be reviewed and replicated, fostering accountability and trust within the scientific community.

Real-World Applications and Success Stories

AI is already demonstrating its potential to drive tangible advancements in scientific research. One notable example is the use of AI to propose a novel hypothesis involving the application of ROCK inhibitors for treating age-related macular degeneration (AMD). This hypothesis, generated through AI analysis, was subsequently tested in wet lab experiments, bridging the gap between theoretical insights and practical applications.

Such success stories highlight the ability of AI to accelerate the pace of discovery by identifying promising research directions that might otherwise go unnoticed. By integrating AI with laboratory work, researchers can streamline the transition from hypothesis generation to experimental validation, ultimately reducing the time required to achieve meaningful results.

Challenges and Limitations of AI in Research

Despite its fantastic potential, AI is not a universal solution to all scientific challenges. Certain bottlenecks, such as the time required for clinical trials or the ethical considerations surrounding experimental research, cannot be resolved by AI alone. Additionally, AI systems may encounter difficulties in scenarios where data is limited, ambiguous, or highly context-dependent, necessitating human judgment and expertise.

Your role as a researcher remains indispensable in guiding AI systems, interpreting their outputs, and making informed decisions. While AI can automate many aspects of the research process, it still relies on human oversight to ensure that its conclusions are accurate, relevant, and aligned with broader scientific goals.

Open Science and Collaborative Innovation

The development of AI in science aligns closely with the principles of open science and collaboration. Open source tools provide widespread access to access to advanced technologies, allowing researchers from diverse backgrounds and institutions to contribute to and benefit from AI-driven discoveries. However, balancing the ideals of open science with the need for intellectual property protection, particularly in fields like biotechnology, remains a complex challenge.

By fostering collaboration while respecting commercial interests, the scientific community can maximize the impact of AI on research. Open science initiatives also promote transparency, allowing researchers to build on each other’s work and accelerate progress. This collaborative approach ensures that the benefits of AI are distributed widely, driving innovation across disciplines and regions.

Shaping the Future of Scientific Discovery

The ultimate vision for AI in research is the creation of a fully integrated virtual laboratory where AI agents collaborate seamlessly to automate complex workflows. Such a system could transform science by eliminating intelligence bottlenecks and allowing faster, more informed discoveries. As AI continues to evolve, its role in hypothesis generation, experimental planning, and data analysis will expand, offering new opportunities to address pressing challenges such as curing diseases, combating climate change, and extending human lifespan.

By embracing the potential of AI while addressing its limitations, researchers can harness this technology to push the boundaries of what is possible in science. The integration of AI into research holds immense promise for tackling some of humanity’s most critical issues, paving the way for a future where scientific discovery is faster, more efficient, and more impactful than ever before.

Media Credit: Freethink

Filed Under: AI, Top News

Latest Geeky Gadgets Deals

Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.

Source link

AI Research

Man leaves Meta to start his own company, now offering ₹17 crore job to…

Published

1 hour ago

September 16, 2025

Ayushmann Chawla

Man leaves Meta to start his own company, now offering ₹17 crore job to… | Hindustan Times (HT Tech)

₹17 crore AI job,AI researcher salary,Meta AI talent war,Memories.ai funding” />
₹17 crore AI job,AI researcher salary,Meta AI talent war,Memories.ai funding” />

Source link

AI Research

Tampere University GPT-Lab hiring doctoral researchers in generative AI

Published

2 hours ago

September 16, 2025

Emma Thompson

Tampere University has announced that GPT-Lab, part of its Computing Sciences Unit, is hiring three to five doctoral researchers in generative AI and software engineering.

The lab works across artificial intelligence, software engineering, and human-computer interaction, combining research and education in Finland and internationally.

The openings were shared in a LinkedIn post by GPT-Lab, which stated: “GPT-Lab (Tampere University) is looking for Doctoral Researchers in Generative AI & Software Engineering to join our team.”

Qualifications highlight AI expertise and development skills

Candidates must hold a master’s degree in computer science, software engineering, data science, artificial intelligence, or a related field. Students close to finishing a master’s by December 2025 may also apply.

The lab says applicants must demonstrate strong written and spoken English. Preferred qualifications include peer-reviewed publications in AI or software engineering, experience in academic or industrial software development, and familiarity with frameworks such as PyTorch, TensorFlow, or Hugging Face.

The recruitment process involves four stages: screening, a video submission, a technical task, and a final interview. Successful candidates must also apply separately for doctoral study rights at Tampere University, as the employment and study admissions are distinct.

Applications must be submitted through Tampere University’s portal by October 3, 2025, at 23:59 Finnish time. Positions are for four years, with a starting salary of €2,714 per month under the Finnish University Salary System.