Why typing is costing you more time than you think

Add it up: how many hours a week do you spend typing, whether it’s charting, emails, applications, or messages? For most clinicians and trainees, the honest answer is a lot, and much of it is done on personal time. And the tool we reach for by default, the keyboard, may be the slower and more taxing option.

Consider the evidence: Bourdin and Fayol (1994) found that the transcription demands of writing, including handwriting and spelling, consume cognitive resources that are not required for speaking. This effect was evident in children, whose writing processes were not yet fully automatized. It was also observed in educated adults under cognitively demanding conditions, with the added transcription demands reducing the quality of the written text they produced (Bourdin & Fayol, 2002).

It’s also slower. A 2017 study compared speech recognition with a mobile touch keyboard on the iPhone and found English speech input was 2.93 times faster (153 vs. 52 wpm) under laboratory conditions (Ruan et al., 2017).

If speaking is both faster and lighter on your mind, the keyboard may not deserve to be your default for the writing you do off the clock, and—with the right precautions—increasingly inside the clinic too.

Start with studying. Verbalizing a concept is active recall in disguise. Brain-dump what you remember on a topic, or rough out flashcard prompts by voice, and your energy goes to the material instead of the keyboard. The same ease helps with the AI tools you already lean on: if ChatGPT, Claude, or Perplexity are part of your week, speaking a long, detailed prompt beats typing one, and you hand the model richer context as you go. Admin is the next obvious win. Thorough emails, CVs, and personal statements are all easier to talk through than to type cold, so speak the rough version and tighten it afterwards. And those ideas that surface mid-commute or between patients and vanish before you can write them down? Dictation catches them the moment they land.

So what has actually changed? AI tools for clinical use generally fall into two types: dictation tools for transcribing speech into EHR fields and scribe tools for generating notes from clinical encounters. The barrier to using either was never just speed; it was ensuring proper data handling. Some tools have now invested in the safeguards that make clinical use possible, but it must be done with caution.

Before any AI tool touches patient data, confirm:

• The tool is HIPAA-compliant in the US, or your local equivalent.
• The vendor meets recognized standards like ISO 27001 and SOC 2 Type II.
• The tool abides by your country's privacy and data storage laws.
• You have obtained approval from your hospital, health system, or clinic to use the tool.
• You have obtained patient consent as required by local laws, health system policies, and guidance from your profession's local regulatory body.
• Ensure patient data is never used to train AI models by enabling a privacy or zero-retention mode, and / or signing a BAA (Business Associate Agreement).
• De-identification is turned on where a tool offers it.

Whether you can actually use an AI tool clinically comes down to your local rules and your organization's approval, not the product alone. Always verify the above before entering any patient information into an AI tool.

Speech recognition has come a long way. Dictation that's faster than typing and lighter on your mental load is no longer just for off-the-clock writing. With the right safeguards in place, it's increasingly something you can bring into your clinical day too. Wispr Flow is one of the tools driving this shift, and it's worth knowing about. Most of all, think about the hours you'd reclaim, and what you'd do with them.