It is not something I do very often, but I did a free trial at an AI site. My thought was that paid AI might be better than the free versions. No, it wasn’t.
The hands are as mangled as in the free versions. In this example the face is rather poor, just look at the eye. I had not paid attention to the feet of people much in other AI images. Here AI clearly misses again.
What really surprised me was a recent search. AI seems to have crept into the search engines. It was a somewhat unusual business I searched for. To my surprise one popped up fairly close to home. I checked the address and could not find the business, so I called the number listed in the search result. It was not a working number. So search engines also “hallucinate”, making up fake results. That can really lead to trouble. When search engines become untrustworthy, we really have entered the twilight zone.
Of course, I will continue to check and see if progress is being made. I will skip searching for AI progress, however.
Addendum
After I published this post this popup came up:
WordPress has joined the AI bandwagon. What do you think of its tag recommendations?
That is my question, “Is AI getting better?” — Is progress being made? So, I tried three image generators with a very simple prompt:
“An artist working in her studio. She is at the easel, paintbrush in hand creating a gorgeous landscape painting.“
Here is wat I got – the images here are rescaled but otherwise just as they came from the AI generators.
Hands are a challenge and still a problem for AI. In the first image the generator side-stepped the issue and showed nothing of the hand. The second one really mangled the artists hand in “classic AI” style. It also managed to break up the paintbrush (?) into floating pieces. The third one almost gets it correctly, one hand is awkward, the other obscure. Faces are pretty good, but that has been the hallmark of artificial intelligence image generators.
Has progress been made? Well, all three got “she” and “her” correct. All show an easel, and all have a pretty nice landscape painting in the works. All are indoors, so even “studio” was interpreted correctly.
I will “hand” it to AI that it is a bit improved since last I tried it. But I also have to tell you that a fourth one begged off and did not produce any image at all – claiming to be working on improving its talents.
In the age of rapid technological advancement, artificial intelligence (AI) has emerged as both a marvel and a menace. While AI promises to revolutionize industries and simplify tasks, its inherent flaws pose significant dangers that demand our attention. One of the most critical issues is AI’s inability to discern its own fallibility, leading to what can only be described as digital hallucinations. These hallucinations result in erroneous answers and misguided outcomes, threatening to blur the lines between reality and fiction.
My experience with AI has highlighted a troubling reality: AI systems lack the capacity to recognize when they are wrong. Unlike humans, who possess the ability to question, doubt, and correct themselves, AI operates within the confines of its programming, incapable of self-correction. This limitation manifests in AI “hallucinating” responses, providing answers that are patently incorrect yet delivered with unwavering confidence.
What’s even more concerning is the propensity of individuals to readily accept these flawed answers without question. In an era where technology reigns supreme, there exists a dangerous tendency to unquestioningly trust AI-generated information. This blind faith in AI not only perpetuates misinformation but also undermines critical thinking and discernment.
The consequences of this blind trust are manifold. From misinformed decision-making in critical sectors such as healthcare and finance to the spread of fake news and propaganda, the dangers of AI’s unchecked influence are evident. Moreover, as AI continues to permeate every aspect of our lives, the risk of confusing reality with fiction becomes increasingly pronounced.
But amidst these perils lies the possibility of a solution. Recognizing that AI mirrors our own susceptibility to gullibility is the first step towards mitigating its dangers. Just as humans must exercise skepticism and critical thinking in the face of information overload, so too must we approach AI with caution and discernment.
Education emerges as a powerful tool in combating the dangers of AI. By fostering a culture of digital literacy and equipping individuals with the skills to evaluate information critically, we can empower society to navigate the complexities of the digital age. Additionally, stringent regulations and ethical guidelines must be established to govern the development and deployment of AI technologies, ensuring accountability and transparency.
Ultimately, the onus lies on us — both as creators and consumers of technology — to harness the potential of AI responsibly. By acknowledging its limitations and actively working towards solutions, we can harness the benefits of AI while safeguarding against its inherent dangers. Only then can we ensure that AI remains a tool for progress rather than a harbinger of chaos.
Visual Deception: Illustrating the Pitfalls of AI Perception
In tandem with the discussion of AI’s pitfalls lies another unsettling truth: the AI’s capacity for visual deception. Despite providing clear instructions and references, AI-generated images often diverge from reality, showcasing a disconcerting inability to accurately depict the intended subject matter.
We all have laughed at the inability of AI to depict human hands accurately. It is in the little details where we can see the limitations and errors.
My illustrations here depict one such inability of AI. I asked an image generator for dogwood blossoms. The results were far from satisfactory. Instead of capturing the delicate beauty of dogwood blossoms, the AI-generated images strayed into a realm of “hallucinations” presenting distorted interpretations that bore little resemblance to actual dogwood blossoms.
Real Dogwood Blossom
Even when presented with an actual photograph of dogwood blossoms, the generator could not create images that faithfully present the actual details correctly.
This visual discrepancy underscores a fundamental flaw in AI’s perception: its propensity for “hallucinating” visuals akin to its cognitive inaccuracies. Just as AI provides erroneous answers when faced with complex queries, it similarly falters in rendering accurate representations of visual stimuli.
The implications of this visual deception extend beyond mere aesthetic concerns. In fields such as design, art, and advertising, where visual accuracy is paramount, AI’s inability to faithfully recreate intended images can lead to significant miscommunications and misunderstandings.
Moreover, the prevalence of AI-generated images in digital media raises ethical questions regarding authenticity and trust. As society grapples with the proliferation of deepfakes and manipulated visuals, the inability of AI to produce reliable images exacerbates the erosion of trust in digital content.
Nevertheless, amidst the sea of visual deception lies an opportunity for reflection and innovation. By acknowledging the limitations of AI in visual perception, we can develop strategies to enhance its accuracy and reliability. This may involve refining algorithms, incorporating human oversight, or implementing robust verification processes to ensure the fidelity of AI-generated images.
Additionally, fostering digital literacy and critical thinking skills becomes imperative in navigating the landscape of AI-generated visuals. Educating individuals on the nuances of image manipulation and encouraging skepticism towards digital content can empower them to discern fact from fiction in an era of visual deception.
As we confront the challenges posed by AI’s visual shortcomings, we must remain vigilant in our quest for solutions. By addressing these issues head-on and striving for transparency and accountability in AI development, we can mitigate the dangers of visual deception and pave the way towards a more trustworthy digital future.
This article was generated by ChatGPT but carefully edited to present my opinions accurately.
Even WordPress incorporates AI. Here is its suggestions upon publication of this article:
A New Medical Building We visited a brand-new medical building. Here are some views of the windows from the outside and one from the inside. … Last week fellow Monday Window bloggers shared these posts: Climb up the stalk St Mary’s Cathedral and its Windows Windows, 1889 white framed window Monday Window Windows (on water) Vintage […]
Window, blind, and shadows Last week was the solar eclipse, however at my location in Georgia it was only a partial one. Rather than going outside and viewing it with appropriate glasses I decided to see a television feed of the total solar eclipse and also project the partial eclipse as a “floor show” inside. […]
.:. © 2024 Ludwig Keck
This 'n That 