The survey Verheggen et al. (2014) found that scientists who work specifically on the physical science basis of climate change (known as WG1 for short), especially experts in attribution and aerosols, tend to believe humans are responsible for a greater proportion of the observed global warming than other scientists working in climate science who may, in some cases, have a "second-order belief" (see figure 2 in that study).

Overall, studies show 91% to 97% consensus among experienced climate scientists.

A one-year increase in Arctic sea ice extent is short-term noise caused by weather, and is not indicative of a long-term recovery from the rapid human-caused decline.

Sea ice extent has dropped since 2013.

Here's what it means: the scientist self-ratings survey included in Cook et al. (2013) found a 96% consensus that humans are responsible for most of the current global warming, among papers that explicitly quantified whether or not humans were responsible. (This was not originally stated in the paper, but can be seen in the anonymized raw data of The Consensus Project.)

The contrarian Roy Spencer did claim to be part of the "97% consensus". The number 97% was produced with different criteria but The Consensus Project database shows that his claim is false. (Judith Curry herself did have two old papers among the 97%, but has published no papers stating whether or not humans cause most of the observed warming.)

Every part of the Earth's climate system has continued warming since 1998.

Global warming apparently slowed down, but did not stop over reasonable averaging periods (more than 10 years). The slowdown is now over. The slowdown may have partly been an illusion produced by 1998 (an unusually hot year), plus a lack of arctic data (coverage bias) during a time when arctic regions were warming quickly. Instrumental bias may also be a factor. Ocean heat content continued rising during the slowdown.