Tag Archives: AI Artificial Intelligence

(comment & excerpts from…) Artificial intelligence research may have hit a dead end

“Misfired” neurons might be a brain feature, not a bug — and that’s something AI research can’t take into account

By THOMAS NAIL
APRIL 30, 2021 10:00PM (UTC)

https://www.salon.com/2021/04/30/why-artificial-intelligence-research-might-be-going-down-a-dead-end/

[…]  artificial intelligence researchers and scientists are busy trying to design “intelligent” software programmed to do specific tasks. There is no time for daydreaming.

Or is there? What if reason and logic are not the source of intelligence, but its product? What if the source of intelligence is more akin to dreaming and play?

Recent research into the “neuroscience of spontaneous fluctuations” points in this direction. If true, it would be a paradigm shift in our understanding of human consciousness. It would also mean that just about all artificial intelligence research is heading in the wrong direction.

Yet all approaches have one thing in common: they treat intelligence computationally, i.e., like a computer with an input and output of information. 

Narrow AI excels at accomplishing specific tasks in a closed system where all possibilities are known. It is not creative and typically breaks down when confronted with novel situations. On the other hand, researchers define “general AI” as the innovative transfer of knowledge from one problem to another.

Decades of neuroscience have experimentally proven that neurons can change their function and firing thresholds, unlike transistors or binary information. It’s called “neuroplasticity,” and computers do not have it.  

Spontaneous fluctuations are neuronal activities that occur in the brain even when no external stimulus or mental behavior correlates to them. These fluctuations make up an astounding 95% of brain activity while conscious thought occupies the remaining 5%. In this way, cognitive fluctuations are like the dark matter or “junk” DNA of the brain. They make up the biggest part of what’s happening but remain mysterious.   

Neuroscientists have known about these unpredictable fluctuations in electrical brain activity since the 1930s, but have not known what to make of them. Typically, scientists have preferred to focus on brain activity that responds to external stimuli and triggers a mental state or physical behavior. They “average out” the rest of the “noise” from the data.

This is why computer engineers, just like many neuroscientists, go to great lengths to filter out “background noise” and “stray” electrical fields from their binary signal. 

This is a big difference between computers and brains. For computers, spontaneous fluctuations create errors that crash the system, while for our brains, it’s a built-in feature.    

What if noise is the new signal? What if these anomalous fluctuations are at the heart of human intelligence, creativity, and consciousness? 

There is no such thing as matter-independent intelligence. Therefore, to have conscious intelligence, scientists would have to integrate AI in a material body that was sensitive and non-deterministically responsive to its anatomy and the world. Its intrinsic fluctuations would collide with those of the world like the diffracting ripples made by pebbles thrown in a pond. In this way, it could learn through experience like all other forms of intelligence without pre-programmed commands. 

In my view, there will be no progress toward human-level AI until researchers stop trying to design computational slaves for capitalism and start taking the genuine source of intelligence seriously: fluctuating electric sheep.

My comment/reflections…

Yes, I read this and excerpted elements that resonated particularly strongly with me. Whenever I hear discussions about AI, I have misgivings. This article helps me to articulate some of these.

Notions such as creativity, addressing ‘novel situations’, going beyond ‘what is known’, or programmed, to find novel solutions that may not have been already attempted. A “closed system where all possibilities are known” is simply a translation of human fallibility with all its potential biases and blind spots, into, as the author says, “computational slaves for capitalism”. One that works faster, cheaper, more efficiently, but without the potential for the fluctuations and ‘noise’ to get in the way.

Well this ‘noise’, to me, is the human condition and I believe it contributes to the wonders of diversity, of difference, of creativity and even what might be considered bohemian or eccentric responses and ways of being that provide the colours of our world.

In terms of the origins of the new technological and AI machinery, what would it mean in terms of the ethics, morals and understandings of ‘right and wrong’, good/bad, acceptability of ‘solutions’, if any nation, sect or belief system of the world was able to program and develop it? Any religion, any philosophy, any group or individual? We know who is ruling the development of AI right now, is that ok with you and me? With our neighbours, our extended families, our region or our place in the world? Have we thought about why this might be, or how it might feel different if our own belief systems were completely incompatible or in opposition?

ARticle review: This Researcher Says AI Is Neither Artificial nor Intelligent

Kate Crawford, who holds positions at USC and Microsoft, says in a new book that even experts working on the technology misunderstand AI. 

TECHNOLOGY COMPANIES LIKE to portray artificial intelligence as a precise and powerful tool for good. Kate Crawford says that mythology is flawed. In her book Atlas of AI, she visits a lithium mine, an Amazon warehouse, and a 19th-century phrenological skull archive to illustrate the natural resources, human sweat, and bad science underpinning some versions of the technology.

book cover - Atlas of AI by Kate Crawford Link to book review.

Crawford, a professor at the University of Southern California and researcher at Microsoft, says many applications and side effects of AI are in urgent need of regulation.

Crawford recently discussed these issues with WIRED senior writer Tom Simonite. An edited [and further excerpted] transcript follows.

KATE CRAWFORD: It [AI] is presented as this ethereal and objective way of making decisions, something that we can plug into everything from teaching kids to deciding who gets bail. But the name is deceptive: AI is neither artificial nor intelligent.

You take on that myth by showing how AI is constructed. Like many industrial processes it turns out to be messy. Some machine learning systems are built with hastily collected data, which can cause problems like face recognition services more error prone on minorities.

We need to look at the nose to tail production of artificial intelligence. The seeds of the data problem were planted in the 1980s, when it became common to use data sets without close knowledge of what was inside, or concern for privacy. It was just “raw” material, reused across thousands of projects.

This evolved into an ideology of mass data extraction, but data isn’t an inert substance—it always brings a context and a politics. 

You trace the roots of emotion recognition software to dubious science funded by the Department of Defense in the 1960s. A recent review of more than 1,000 research papers found no evidence a person’s emotions can be reliably inferred from their face.

Emotion detection represents the fantasy that technology will finally answer questions that we have about human nature that are not technical questions at all. This idea that’s so contested in the field of psychology made the jump into machine learning because it is a simple theory that fits the tools. Recording people’s faces and correlating that to simple, predefined, emotional states works with machine learning—if you drop culture and context and that you might change the way you look and feel hundreds of times a day

We’ve seen research focused too narrowly on technical fixes and narrow mathematical approaches to bias, rather than a wider-lensed view of how these systems integrate with complex and high stakes social institutions like criminal justice, education, and health care. I would love to see research focus less on questions of ethics and more on questions of power. These systems are being used by powerful interests who already represent the most privileged in the world.

Is AI still useful?

Let’s be clear: Statistical prediction is incredibly useful; so is an Excel spreadsheet. But it comes with its own logic, its own politics, its own ideologies that people are rarely made aware of.

https://www.wired.com/story/researcher-says-ai-not-artificial-intelligent/

(My highlighting) Highlighted parts relate directly to my thinking in regards to how AI/technology can be used across a general (diverse) population, when it has been designed and programmed by fallible and inevitably biased humans? As fashions change, theory, perspectives, experiences, culture/s, languages and dialects, and effects of globalisation, first world power and dominance, disparities between the global ‘North and South’, the ‘East and West’, religious and political influence, AI is being built and programmed by who? As the author says in the final comment, AI “comes with its own logic, its own politics, its own ideologies that people are rarely made aware of” and this is one of my main concerns. How can this be mitigated? Should we (users/educators) be cognisant of these issues of power and bias when we chose our tools? Should we ensure we educate our learners to be critical, to always consider minority perspectives, to consider the tools they/we use for what might be missed, or not considered, or how they support and ensure the power (and knowledge) is wielded by those with conflicting interests?

A Leve Reflections: 1 May, 2021