In March 2022 Gary Marcus published an article arguing that deep learning is hitting a wall.

True if Gary Marcus published an article in March 2022 that argues deep learning is hitting a wall (for example one titled or framed as 'Deep Learning Is Hitting a Wall'), as shown by the publication's dated record. Mostly true if the article was published within a few days outside March 2022 (e.g. late February or early April) and makes that argument. Mixed if it was published in 2022 but more than a month from March, or the 'hitting a wall' argument is only a side point. Mostly false if such an article exists but was published in another year. False if Marcus published no article making that argument around that time.

Set before the research began.

1. E1 Deep Learning Is Hitting a Wall - Nautilus Nautilus · 10 March 2022

   “Indeed, we may already be running into scaling limits in deep learning, perhaps already approaching a point of diminishing returns.”

2. E2 Deep Learning Is Hitting a Wall - Nautilus Nautilus · 10 March 2022

   “Gary Marcus is a scientist, best-selling author, and entrepreneur.”

3. E3 Gilbane Advisor — neurosymbolic hybrid, content understanding Gilbane Advisor (Frank Gilbane) · 30 March 2022

   “Gary Marcus provides historical and recent context and makes a case for a neurosymbolic hybrid approach.”

4. E4 Has Sam Altman gone full Gary Marcus? Gary Marcus (Substack) · 16 November 2023 Own record

   “On March 20, 2022 in Nautilus, in the essay I am perhaps most notorious for, I suggested that deep learning might be approaching a wall”

The Nautilus essay "Deep Learning Is Hitting a Wall" carries a publication date of 2022-03-10, which falls in March 2022 (E1). The essay argues that deep learning may be reaching scaling limits and diminishing returns (E1). Its author biography names Gary Marcus (E2). A Gilbane Advisor roundup dated 2022-03-30 independently mentions a Marcus essay from that period that makes the case for a neurosymbolic hybrid approach (E3). Marcus's own later post (E4) gives a slightly different date but agrees it was published in March 2022.