Category Archives: Artificial Intelligence

More on AI Hallucinations…

Posted on by
Reply

The mainstream adoption of AI continues to reveal the precarious balance between the benefits and the pitfalls.

Yes, AI tools can reduce the time it takes to research information, to draft documents and to complete repetitive tasks.

But, AI is not so good at navigating subtle nuances, interpreting specific context or understanding satire or irony. In short, AI cannot “read the room” based on a few prompts and a collection of databases.

And then there is the issue of copyright licensing and other IP rights associated with the original content that large language models are trained on.

One of the biggest challenges to AI’s credibility is the frequent generation of “hallucinations” – false or misleading results that can populate even the most benign of search queries. I have commented previously on whether these errors are deliberate mistakes, an attempt at risk limitation (disclaimers), or a way of training AI tools on human users. (“Spot the deliberate mistake!) Or a get-out clause if we are stupid enough to rely on a dodgy AI summary!

With the proliferation of AI-generated results (“overviews”) in basic search queries, there is a tendency for AI tools to conflate or synthesize multiple sources and perspectives into a single “true” definition – often without authority or verified citations.

A recent example was a senior criminal barrister in Australia who submitted fake case citations and imaginary speeches in support of a client’s case.

Leaving aside the blatant dereliction of professional standards and the lapse in duty of care towards a client, this example of AI hallucinations within the context of legal proceedings is remarkable on a number of levels.

First, legal documents (statutes, law reports, secondary legislation, precedents, pleadings, contracts, witness statements, court transcripts, etc.) are highly structured and very specific as to their formal citations. (Having obtained an LLB degree, served as a paralegal for 5 years, and worked in legal publishing for more than 10 years, I am very aware of the risks of an incorrect citation or use of an inappropriate decision in support of a legal argument!!!)

Second, the legal profession has traditionally been at the forefront in the adoption and implementation of new technology. Whether this is the early use of on-line searches for case reports, database creation for managing document precedents, the use of practice and case management software, and the development of decision-trees to evaluate the potential success of client pleadings, lawyers have been at the vanguard of these innovations.

Third, a simple document review process (akin to a spell-check) should have exposed the erroneous case citations. The failure to do so reveals a level laziness or disregard that in another profession (e.g., medical, electrical, engineering) could give rise to a claim for negligence. (There are several established resources in this field, so this apparent omission or oversight is frankly embarrassing: https://libraryguides.griffith.edu.au/Law/case-citators, https://guides.sl.nsw.gov.au/case_law/case-citators, https://deakin.libguides.com/case-law/case-citators)

In short, as we continue to rely on AI tools, unless we apply due diligence to these applications or remain vigilant to their fallibility, we use them at our peril.

 

Pop in Perpetuity

Posted on by
Reply

Exactly a year ago, I blogged about ageing rockers and their propensity to continue touring and recording. This past weekend I experienced two events that provided almost polar opposites as to how musicians will perpetuate their “live” legacy. (Of course, in theory, their recordings will last forever, in physical, digital and streaming formats – as long as the equipment, technology and platforms survive…)

On the one hand, there was the Sun Ra Arkestra, who since their founder’s death in 1993, have continued to play the music of Sun Ra, respecting the sound, format and spirit of the original band formed in the 1950s. Some of the current band members played with Sun Ra himself, so there is a thread of continuity that connects us back to the past. But even as these surviving members depart this world, the music of Sun Ra will live on in concert form through subsequent generations of players. This type of perpetuity is not uncommon among bands of the 60s, 70s and 80s, although in many cases, there is usually at least one original band member performing, or members who overlapped with the band founders. (Some notable exceptions: Soft Machine, who continue performing and recording, but whose remaining original member left nearly 50 years ago; and Faust, who split into at least two separate bands that still tour and record under the same name.)

On the other hand, there was the high-tech concert presentation by the late composer and performer Ryuichi Sakamoto, entitled KAGAMI. This involved the use of AR headsets and a 3D avatar of Sakamoto, captured in sound and vision performing a selection of his music, sat at a grand piano. The audience, initially seated in a circle around the virtual performance area in order to acclimatise to what they were seeing, was invited to move around the avatar, and even peer into the open grand piano. Two things were striking: first, the 360 degree image was very impressive in the level of detail; second, even if someone was standing between the viewer and the avatar zone, the headset still presented the image of Sakamoto sat at the keyboard. The technology not only captures a digital visualisation of the pianist in action, it also replicates the notes he played as well as the tonal expression and the timbres, resonances and acoustics of the physical instrument. While the audio HiFi was superior to the atavistic CGI, the latter will no doubt improve; as will the slightly clunky and heavy headsets – the 50 minute duration is probably the most I could have endured.

Neither format of the above concerts is better or superior to the other. Both are authentic in their own way, and true to the artistry of musicians they celebrate. Of course, if we end up using AI to compose “new” music by Sakamoto, that may undermine that authenticity. But given Sun Ra’s origin story, I wouldn’t be surprised if he started beaming his new works from Saturn.

 

A postscript on AI

Posted on by
Reply

AI tools and related search engines should know when a factual reference is incorrect, or indeed whether an individual (especially someone notable) is living or dead. In an interesting postscript to my recent series on AI, I came across this article – written by someone whom Google declared is no longer with us.

Glaring errors like these demand that tech companies (as well as publishers and media outlets who increasingly rely on these tools) take more seriously the individual’s right of reply, the right to correct or amend the record, as well as the right to privacy and to be forgotten on the internet.

As I commented in my series of articles, AI tools such as ChatGPT (and, it seems, Google Search) can easily conflate separate facts into false statements. Another reason to be on our guard as we embrace (and rely on) these new applications.

Next week: Bad Sports

 

 

Some final thoughts on AI

Posted on by
Reply

Last week, I attended a talk by musical polymath, Jim O’Rourke, on the Serge Paperface modular synthesizer. It was part memoir, part demonstration, and part philosophy tutorial. At its heart, the Serge is an outwardly human-controlled electronic instrument, incorporating any number and combination of processors, switches, circuits, rheostats, filters, voltage controllers and patch cables. These circuits take their lead from the operator’s initial instructions, but then use that data (voltage values) to generate and manipulate sound. As the sound evolves, the “composition” takes on the appearance of a neural network as the signal is re-patched to and from each component, sometimes with random and unexpected results – rather like our own thought patterns.

But the Serge is not an example of Artificial Intelligence, despite its ability to process multiple data points (sequentially, in parallel, and simultaneously) and notwithstanding the level of unpredictability. On the other hand, that unpredictability may make it more “human” than AI.

My reasons for using the Serge as the beginning of this concluding blog on AI are three-fold:

First, these modular synthesizers only became viable with the availability of transistors and integrated circuits that replaced the valves of old, just as today’s portable computers rely on silicon chips and microprocessors. Likewise, although some elements of AI have been around for decades, the exponential rise of mobile devices, the internet, cloud computing and social media has allowed AI to ride on the back of their growth and into our lives.

Second, O’Rourke referred to the Serge as being “a way of life”, in that it leads users to think differently about music, to adopt an open mind towards the notion of composition, and to experiment knowing the results will be unpredictable, even unstable. In other words, suspend all pre-conception and embrace its whims (even surrender to its charms). Which is what many optimists would have us do with AI – although I think that there are still too many current concerns (and the potential for great harm) before we can get fully comfortable with what AI is doing, even if much of may actually be positive and beneficial. At least the Serge can be turned off with the flick of a switch if things get out of hand.

Third, as part of his presentation O’Rourke made reference to Stephen Levy’s book, “Artificial Life”, published 30 years ago. In fact, he cited it almost as a counterfoil to AI, in that Levy was exploring the interface between biological life and digital DNA in a pre-internet era, yet his thesis is even more relevant as AI neural nets become a reality.

So, where do I think we are in the evolution of AI? A number of cliches come to mind – the Genie is already out of the bottle, and like King Canute we can’t turn back the tide, but like the Sorceror’s Apprentice maybe we shouldn’t meddle with something we don’t understand. I still believe the risks associated with deep fakes, AI hallucinations and other factual errors that will inevitably be repeated and replicated without a thought to correct the record represent a major concern. I also think more transparency is needed as to how LLMs are built, and the way AI is trained on them, as well as disclosures when AI is actually being deployed, and what content has been used to generate the results. Issues of copyright theft and IP infringements are probably manageable with a combination of technology, industry goodwill and legal common sense. Subject to those legal clarifications, questions about what is “real” or original and what is “fake” or artificial in terms of creativity will probably come down to personal taste and aesthetics. But expect to see lots of disputes in the field of arts and entertainment when it comes to annual awards and recognition for creativity and originality!

At times, I can see AI is simply a combination of mega databases, powerful search engines, predictive tools, programmable logic, smart decision trees, pattern recognition on steroids, all aided by hi-speed computer processing and widespread data distribution. At other times, it feels like we are all being made the subject matter or inputs of AI (it is happening “to” us, rather than working for us), and in return we get a mix of computer-generated outputs with a high dose of AI “dramatic license”.

My over-arching conclusion at this point in the AI journey is that it resembles GMO crops – unless you live off grid and all your computers are air-locked, then every device, network and database you interact with has been trained on, touched by or tainted with AI. It’s inevitable and unavoidable.

Next week: RWAs and the next phase of tokenisation