Foundations of mind 5: The new AI scare
Room 304 CIIS • 1453 Mission St, San Francisco, CA 94103 Nov 3-4 2017
In terms of downloads (150k per year), annual page views (27 million +) and peer-reviewed papers ( over 100 in its first 3 years from March 2014 to March 2017), Foundations of mind (http://foundationsofmind.org/) is now the world’s leading science of mind research group. While centered on cognitive science, it has featured many papers on the quantum mechanics view of mind, the foundations of physics and biology, and indeed ecology and health as manifestations of mind.
Its most recent proceedings volume published in March 2017 received a total of 4,333 downloads in its first month, with the top papers receiving 750+, about what ACM papers typically take 25 years to achieve. Please note we are secular, and totally independent; we have never taken corporate or state money, nor have we gamed the system as others do by asking our students to link to or download our work.
Abstracts to firstname.lastname@example.org by July 1 2017. Invitation to present at conference and submit a full paper will follow by July 8. Full papers by July 31; notification of acceptance for the proceedings volume will follow your presentation at the conference.
Nov 3: The new AI scare
AI systems will keep us as pets. Massive unemployment as computers learn to drive cars, translate accurately, stock shelves, and much else. Silicon-based life forms reflecting on the brief carbon-based intelligence era before the singularity.
All of these themes are rampant in the second decade of the 21st century. Yet they do not reveal the full story. The “pets” idea, repeated ad nauseam by Elon Musk, is an echo of Marvin Minsky a generation ago. It is a safe bet that no, you will not be ordering a driverless taxi to take you to a random destination you specify anytime before 2022. The much-vaunted successes of Deep Thought are due to a venerable algorithm called stochastic gradient descent, which now converges to a solution because computers are about 100 million times faster than they were during the last AI scare of the 1980’s and almost infinitely faster than during Minsky’s abortive 1950’s scare.
This session invites papers on what computers, even now, can’t do and above all why that is the case. It encourages participants to speculate on aspects of the architecture of the cortex that allows a 20 Watt “computer” (that would be the thing under your skull) to outperform machines with Terahertz speed and petabyte memory. It invites papers on quantum computation as a natural outcome of human cognition and noesis. It also welcomes those who in turn wish to usher in the new post-singularity era. After all, the phrase and concept originally derive from Johnny von Neumann, the founder of many aspects of computing as well as quantum mechanics.
Above all, we invite recapitulation of the themes broached in Penrose’s masterpiece about the relationship between consciousness and computability. His viewpoint fused elements from the theory of computability (partially recursive functions), non-determinism, tiling theory and phenomenology. The relationship with formal language theory was not explored, nor was an explicit engagement with the fact that all observer-dependent measurement in QM involves the infinite. The re-assertion of human noesis may pave the way for that of human dignity.
Specifically, we invite papers on the following themes;
- Machine Translation (MT). Computers have been getting MT almost right since the 1950’s. attendees are asked to consider two dilemmas; that checking a translation is at least as time-consuming as producing one, and that 99.9% correct is not good enough even for an academic textbook.
- Kill Chain; use of AI systems in the “Kill Chain” in combat is burgeoning. Yet it is being done in the face of objections like military strategists Rivollo and van Riper, who insist on precise targeting and reading of the enemy’s psychology.
- Neglected research: the apparent magic wrought by the likes of Sutskever at Google has meant that linguistics research into issues like speech-acts and conversational implicatures is being elided? The inappropriate use of the word “neural” as a modifier for convolutional nets has led to neuroscience reverting to its 1940’s archetype with invasive fiddling around under the hood again the method de jour?
- Language was famously promoted as the royal road to the infinite by the symbolists. What is the relation between their “semantic transcendentalism” and the role of infinite quantities in quantum mechanics? What does this mean for AI?
- Finally, we invite papers that explore the fact that ordering on a smartphone has met that both AirBnB and Uber were somehow exempt from normal regulations for accommodation and transport. Could this AI scare be a way of persuading us that resistance is futile in our Taskrabbit dystopia? Papers also are welcome that explain that it is in fact a Utopia. Or is it arguable that the successes of Google, Facbook et al are largely based on truly cosmic levels of copyright violation, and that states should administer many of their functions under the rubric of public libraries (and indeed e-mail systems)
Papers also are invited on the classic foundations of mind themes; the formal inadequacy of received accounts of mind and brain (leading to $6 billion being spent by the NIH in 25 years on brain imaging without a single therapy emerging); the foundations of biology; the observer in physics; quantum mind..