Two Cheers for Rebooting AI: Building Artificial Intelligence We Can Trust

By Stephan Kolassa. 2020

364 citations recorded by the source corpus.

Subfields: Alignment & RLHF, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Alignment & RLHF

In Safety, Robustness & Interpretability

Citations in the corpus

Experience Grounds Language (2020, neutral): ““deceit and glamour” (Pierce, 1969) to complaints of humanity’s “gullibility gap” (Marcus and Davis, 2019).”
AGIEval: A Human-Centric Benchmark for Evaluating Foundation Models (2023, neutral): “However, despite their impressive performance on various benchmarks, concerns have been raised about the reasoning abilities, trust-fulness and real-world applicability of these models (Marcus and Davis, 2019).”
The Next Decade in AI: Four Steps Towards Robust Artificial Intelligence (2020, background): “I thank both Yoshua and Vince for making that possible I also thank Deen Abiola, Doug Bemis, Emily Bender, Vince Boucher, Ernie Davis, Tom Dietterich, Pedro Domingos, Chaz Firestone, Artur D'Avila Garcez, Daniel Kahneman, Katia Karpenko, Kristian Kersting, Luis Lamb, Adam Marblestone, Melanie Mitche”
Selection-Inference: Exploiting Large Language Models for Interpretable Logical Reasoning (2022, background): “It is also one of the most valued strengths of classical, symbolic AI over contemporary deep learning methods (Bengio et al., 2021; Marcus, 2020; Marcus and Davis, 2019), prompting the recent increase in the use of neurosymbolic approaches to bridge this gap (Garcez and Lamb, 2020; Garnelo and…”
Word meaning in minds and machines (2020, background): “However, BERT ’ s predictions are sensitive to small changes in wording — a property of current NLP systems that has been observed elsewhere (Jia & Liang, 2017; Marcus & Davis, 2019).”
Omnidirectional Transfer for Quasilinear Lifelong Learning (2021, background): “This inability to omnidirectionally transfer has been identified as one of the key obstacles limiting the capabilities of artificial intelligence [19, 20].”
Neurosymbolic Programming (2021, background): “At the same time, concerns remain about the use of deep neural networks in real-world problem domains (Marcus and Davis, 2019).”
Digital technologies, innovation, and skills: emerging trajectories and challenges (2021, background): “By paying for the advertisement, is the firm using AI? As with all intangibles, if AI is embedded in software, but is not a “physical thing,” then it is difficult to measure its adoption, even though it may be substantial (Marcus and Davis, 2019).”
Trustworthy Artificial Intelligence: A Review (2022, background): “Marcus and Davis [130] provide an excellent example of facial recognition software that explains the need for trustworthy AI.”
Towards Question Format Independent Numerical Reasoning: A Set of Prerequisite Tasks (2020, methodology): “This was emphasized when the Winograd schema challenge was proposed [Levesque et al. , 2012] and Marcus & Davis further emphasize this in their recent book [Marcus and Davis, 2019].”
The ethics of algorithms: key problems and solutions (2020, neutral): “This is why it is crucial to ensure that data fed to algorithms are validated independently, and data retention and reproducibility measures are in place to mitigate inconclusive evidence leading to unjustified actions, along with auditing processes to identify unfair outcomes and unintended consequ”
A very preliminary analysis of DALL-E 2 (2022, neutral): “However, it is reasonable to question whether DALL-E 2 constitutes progress toward solving the deep challenges of commonsense reasoning, comprehension, reliability, and so forth that would be needed for a truly general-purpose AI (Marcus & Davis, 2019).”
Neurosymbolic Program Synthesis (neutral): “Neurosymbolic machine learning [1,2,3] is an increasingly prominent response to these concerns.”