I use one mainly for monitoring fermenting temperature on my carboy (shoot the side of the carboy). Mine has a ±4 °F accuracy which is good enough for me at this time considering it's non-invasive and I'm generally just monitoring temps. I have found that it works reasonably well for a surface temperature. Most times I would guess that it's within 2°F of what it's reading, especially when the device is used in a room-like temperature setting. I usually take temperatures of surfaces around the carboy prior to getting the carboy temperature which I feel helps me decide on the accuracy of the carboy. I have good confidence in it's readings most times.
Where I HAVE seen poor readings are:
1) Battery getting low and readings start bouncing around a bit.
2) If I take it in a cold environment (<~50°F) for several minutes it will begin to read lower than actual by many degrees.
3) It cannot read surface temperature on stainless surfaces - totally inconsistent/incorrect readings.
It can be used in a liquid scenario decently well if you swirl up the liquid to get the center liquid to the surface and you keep your laser on the up-swirling liquid area. I use it in candy making a lot and it works really well. I recently tested my instant read digital thermometer, coarse meat thermometer, analog mercury thermometer, and infrared thermometer against each other in a cup of hot water. The results were that the instant-read and infrared were within a degree of each other, the meat thermometer was in about the same range, but the mercury was about 10°F low. My conclusion was that the mercury thermometer was not accurate.