What did the 2020 report predict?

The pandemic-year forecasts covered model scale, vision transformers, drug discovery, defense startups, and the proposed NVIDIA-Arm deal. It published 8 forecasts for the next twelve months; the complete original list appears on the predictions topic page.

Are these predictions graded on this page?

No. This archive preserves the original forecasts. Later assessments are linked separately through the predictions scorecard.

How large did the report predict language models would become?

It predicted the first ten-trillion-parameter model within its forecast horizon. That was a claim about future scale, not a description of GPT-3’s size in 2020.

What shift did the report predict for attention-based networks?

It predicted that attention-based networks would move from NLP into state-of-the-art computer-vision results. The claim concerned transfer of an architectural approach across domains.

What did the defense-startup forecast specify?

It predicted more than $100 million raised collectively by a wave of Chinese and European defense-focused AI startups over twelve months, in response to US defense activity and investment.

What would satisfy the AI drug-discovery financing forecast?

The report predicted an IPO or an acquisition above $1 billion for a leading AI-first drug-discovery startup. It named examples but specified a financing or ownership event, not clinical approval.

What did the report predict about NVIDIA and Arm?

It predicted that NVIDIA would not complete its proposed acquisition of Arm. The forecast concerned the outcome of a particular transaction.

Were all eight predictions about model performance?

No. The list also covered a corporate lab closure, defense financing, a drug-discovery company transaction, and the NVIDIA-Arm acquisition. It combined technical and industry outcomes.

The 2020 forecasts, in their original form

The original conditions define the call

The report published 8 predictions for the next twelve months. Several combined a technical or commercial development with a threshold, deadline, or named group of organizations. Those details matter: an event that meets only part of a prediction should not silently count as the whole outcome.

Keep forecasts separate from later grading

The complete list below preserves the wording of the original report, with PDF line breaks removed. The predictions scorecard provides subsequent assessment separately. This archive keeps the original publication-time claim visible so readers can compare it with later evidence.

A forecast can be specific without being certain

The 2020 list named technologies, organizations, financing thresholds, and a proposed acquisition. Those details made the predictions easier to assess later. They should remain visible even when subsequent events make a forecast look unusually prescient or clearly mistaken.

The eight forecasts, as published

The race to build larger language models continues and we see the first 10 trillion parameter model.
Attention-based neural networks move from NLP to computer vision in achieving state of the art results.
A major corporate AI lab shuts down as its parent company changes strategy.
In response to US DoD activity and investment in US based military AI startups, a wave of Chinese and European defense-focused AI startups collectively raise over $100M in the next 12 months.
One of the leading AI-first drug discovery startups (e.g. Recursion, Exscientia) either IPOs or is acquired for over $1B.
DeepMind makes a major breakthrough in structural biology and drug discovery beyond AlphaFold.
Facebook makes a major breakthrough in augmented and virtual reality with 3D computer vision.
NVIDIA does not end up completing its acquisition of Arm.

2020 report, PDF page 172. See the predictions scorecard for subsequent grading.

Measure	Reported value	Definition and source
Number of original forecasts	8	Count of the numbered predictions on the source slide.2020 report, PDF page 172
Forecast horizon	Next 12 months	Horizon stated on the original prediction slide.2020 report, PDF page 172