Finance edition
Muse Wonder does well at financial reporting. It is less successful with day-to-day bookkeeping.
Muse Wonder helps with reporting, less so with bookkeeping
Meta’s new model scores well on an accounting test, but results vary by task.
Meta released Muse Wonder on 12 October. The model was tested against 57 AI models. It completed 83.2 per cent of the tasks correctly, tying for second place. It did particularly well at reporting. This includes calculating cash flow (how much money comes in and goes out) and preparing financial statements. It made more mistakes with day-to-day transactions. For example, Muse Wonder made errors with some unpaid bills. Results varied for closing entries and difficult classifications. This difference matters in financial work. A high score for reporting does not mean the model also handles entries well. The test compares different accounting tasks. In repeated tests, Muse Wonder almost always gave the right answer to reporting questions. Most errors were with day-to-day transactions and difficult cases. An overall score does not show which tasks it handles well and which it does not. When Meta released the model, the company had not yet published a safety report.
Background for subscribersThe reports show that an AI model can do well at one task and make mistakes at another. Muse Wonder scored highly on reporting. It struggled with some payments and entries. So an overall score does not tell you enough. Look at results for each task, and have an employee approve changes first. Also record who can access the accounting data and who approves a change. If you use AI on a larger scale, track the costs, oversight and returns too. Greater use does not automatically bring greater value to the business.
Dutch organisations get less value from AI
A survey that takes place every quarter questioned more than 2,100 business leaders. In the second quarter, 82 per cent of Dutch organisations said they were getting real business value from AI. In the third quarter, that figure was 68 per cent. Only 4 per cent say they are seeing measurable business results and growth. Confidence that organisations can manage the risks also fell as they use AI more. The survey shows that greater use does not automatically bring greater results. According to KPMG, attention is shifting from introducing AI to setting it up properly. This means giving AI a regular place in day-to-day work. A few things have become more important: clear responsibility, oversight, and a view of costs and returns.
Background for subscribersThis is a taster.
Subscribers read the whole Finance edition: every story, each with its background in plain language.
Subscribe →Not ready for that? Read the free letter first →
AI and tech, made simple.
This is the Finance edition. Subscribers get the background to every story, in plain language.
Subscribe
inklaretaal · Amsterdam · © 2026