Kerala Technology
Internet staring down an AI abyss

Videos altered by using AI programs have started appearing on the internet and experts warn of a flood of such morphed material. Image courtesy: Twitter

Internet staring down an AI abyss

Hari Kumar By Hari Kumar, on July 03, 2023
Hari Kumar By Hari Kumar, on July 03, 2023

A clip showing Malayalam megastars Mohanlal and Mammootty portraying characters from Francis Ford Coppola's iconic movie “The Godfather” recently went viral on social media. This was followed by Chief Minister Pinarayi Vijayan’s face pasted on a James Bond movie clip, and Trivandrum MP Shashi Tharoor in a “Mission: Impossible” movie scene. Wonder what will come next.

The quality of these deepfakes was decent, and the clips were harmless, even enjoyable. However, the mea culpa uploaded by the individual who apparently created the video, Tom Antony, was unsettling.

“You can create videos like this even if you only have a still photo,” the YouTuber from Kerala said in a video, adding that some who contacted him wanted to know how to make such videos.

While these revelations are nothing new to those following the rapid growth of artificial intelligence (AI), what makes things scary is that any novice can produce such damaging content now. There are programs that can perform such tasks for even the most inexperienced users. In fact, there are websites that offer to put provided pictures onto selected videos, and the ever-improving quality of deepfakes will make them look genuine.

Similar instances have affected the music industry, where synthesised voices of singers were used to create new songs without the knowledge of the original artists. In April, a song called “Heart on My Sleeve” went viral, featuring the voices of Drake and The Weeknd, who were completely unaware of its existence.

At a panel discussion in Los Angeles,, the frontman of the Blackeyed Peas, predicted that AI would lead to the release of songs long after the death of singers. Stars like Beyoncé and Taylor Swift would never truly die, he said. This vision became a reality when Paul McCartney recently announced the release of a new song by The Beatles, where AI was used to recover John Lennon’s voice from an old audio tape with the consent of Lennon's estate.

The use of synthetic voice becomes significantly more harmful when it falls into the hands of criminals. There have been reports of people receiving phone calls purportedly from their loved ones, asking for urgent financial assistance and sensitive banking details. Authorities in many countries have issued warnings, urging the general public to be cautious regarding unusual phone calls, even from their family members.

Regenerative AI has already been used to produce fake scientific papers and manipulate legal history, successfully deceiving experts in these fields. The case of a New York lawyer who had to apologise to the court and pay a fine for submitting fabricated litigation examples serves as a stark warning to content generators.

Profit-driven individuals have already exploited regenerative AI. A report in Vice magazine by Jules Roscoe reveals that AI-generated trash books dominated Kindle Unlimited’s young adult romance bestseller list last week. These books were pushed to the top by hacking the bestseller list with AI bots, displacing works written by actual authors.

While Amazon eventually removed the junk books from the top list, many of them still remained available for sale. Blogger and AI analyst Alberto Romero describes some of these AI-generated books as complete rubbish, demonstrating that those who created them didn’t even attempt to make them coherent.

“You may argue that there is already an abundance of low-quality content on the web. True, but just imagine what will happen if the creators of that low-quality content increase their output tenfold, a hundredfold, or even a thousandfold,” Romero warns.

But what are the long-term effects of all this? There are already dire warnings, with notable figures like Geoffrey Hinton and Elon Musk stating that AI poses an existential threat to humanity. Others dismiss this as fearmongering, and AI expert Andrew Ng compares it to “talking about overpopulation on Mars”. Nevertheless, most agree that AI will cause significant disruptions to the current usage of the internet.

Initially, AI chatbots, with their ability to provide precise summaries from vast amounts of data, were seen as a potential threat to Google and its search engine monopoly. Now some AI experts, such as Jack Clark, cofounder of Anthropic and chair of the AI Index at Stanford University, argue that powerful intelligence platforms are damaging the internet itself.

Clark refers to a recent study by researchers from the University of Oxford, University of Cambridge, University of Toronto, and Imperial College London, which indicates that AI platforms are being trained on internet data increasingly filled with text generated by AI. This synthetic data has the potential to undermine AI models, rendering their output unreliable and unusable, he says.

Furthermore, as this synthetic data finds its way into platforms like Wikipedia and mass media organisations publish AI-generated content without proper verification due to its seeming authenticity, the significance of the internet itself could be diminished.

When the information superhighway connecting computers worldwide became a reality, it brought knowledge to the fingertips of millions. Prior to the internet age, sifting through documents and spending hours in libraries was the only way to access detailed information on any subject. Searching the internet (or “Googling” as it later became known) became the go-to method for obtaining information, whether it was about Albert Einstein’s theories or a recipe for making an omelet.

The search for information on products and services eventually led to the internet economy, spurring global business as producers found buyers beyond their borders. The next paradigm shift is currently underway with the AI revolution, poised to revolutionise nearly every aspect of the cyber world.

With the rise of AI, the internet is becoming overwhelmed, according to a report in The Verge magazine. AI platforms produce an abundance of text and images, and soon music and video as well. Their output has the potential to overrun or outcompete the platforms we rely on for news, information, and entertainment, it said.

“This potential damage could affect vast areas of the web that most of us find useful – from product reviews to recipe blogs, hobbyist homepages, news outlets, and wikis,” warns the report.

Doctored images and untruthful text circulated through social media have already altered narratives in many societies, accelerating dangerous levels of polarisation. Now, with industrial-scale production and distribution made possible, such campaigns could undermine the credibility of the internet as anything you read, hear or see on it become suspicious. This could make the internet collapse under the weight of invalidated data.

In Indian mythology, the tale of Bhasmasura describes the rise of a powerful demon lord who gains the ability to turn anything and anyone into ashes by pointing his finger at them. Ultimately, Mahavishnu assumes his female avatar as Mohini and tricks Bhasmasura into dancing to his death, causing the demon to point his finger at himself.

Greek mythology also contains a similar tale with King Midas, who obtains the power to turn everything he touches into gold but quickly realises the detrimental consequences of overwhelming power and gives up the boon.

The power of AI is now enticed by the temptress called money, and the companies behind it see only gold everywhere. Whether AI ends up like the demon king in Indian mythology or the greedy ruler in Greek mythology remains to be seen.

Or perhaps we should ask ChatGPT.



Kerala also feeling the funding winter sniffles

The chill of the funding winter is being felt by Indian startups, and Kerala is no exception. According to some back of envelop calculations done by Robin Panicker, a venture partner at Unicorn India Ventures, total funding in the state has decreased by 89 percent, and the number of deals has fallen by approximately 47 percent. In fact, the sluggishness has been noticeable since the second quarter of the previous year, with only 52.68 million US dollars worth of funding was secured in the last six months of 2022, compared to 196.65 million in the first half.

Overall, Indian startups managed to secure only funds worth 6 billion US dollars in the first half of 2023, a sharp decline considering that the sector received 20 billion in the first half of the previous year, as reported by Entracker. In a detailed breakdown published by the magazine, it also notes that the last unicorn to emerge in India was in September 2022 when healthtech firm Molbio raised funds. The top city for funding remained Bengaluru, with 246 startups receiving funds worth 3 billion US dollars, while Delhi, Mumbai, Pune, and Chennai made up the rest of the top five list.



Melbourne firm acquires Trivandrum startup

The first half of 2023 was also devoid of any blockbuster mergers and acquisitions, although the country recorded 81 such deals. One of the latest acquisitions was the takeover of Trivandrum-based ManMech Smart Solutions by Centelon Solutions, which is headquartered in Melbourne. Founded in 2020, ManMech specialises in software that enabled IoT solutions for businesses and robotics. According to Ajit Stephen, CEO of Centelon Solutions, the synergies between the two firms will stimulate more innovations and growth for the company. Centelon offers IT and data management solutions to businesses and has offices in Pune, Singapore, and the US, with over 60 clients, as stated on their website.



Startup to help students bound for Australia

More news from Australia: Two Indian students, Devarshi Desai and Kunal Sittani, have managed to secure undisclosed seed funding after setting up a forum to assist Indian students looking to study Down Under. Both co-founders state that they faced a lot of difficulties when they arrived in Australia, and this experience motivated them to establish Studynash, a platform to help aspiring students. It has now grown into one of the largest online communities for international students, with 113,000 YouTube subscribers, 29,000 Facebook members, and 21,000 Instagram followers, reports Startupdaily. It provides access to all courses offered at 42 universities across Australia and has established 10 partnerships with private education institutes.



Add a bit of kimchi to your life

Feeling burnt out? A South Korean entrepreneur has a place just for you to recharge. For a small fee, attendees selected after going through a multi-stage interview process can live in a village for six weeks, exploring different interests, getting to know new people, and relaxing in the countryside, reports Quartz. Hong Doo-woo started this enterprise after making money from a mobility startup. The idea is for youngsters to find a place to escape from the daily pressures that start from school days and continue into adulthood. The founder of the Don’t Worry Village says it is also a way to get society to recognise that failure is not a lifetime sentence and to make businessmen and job seekers realise that a setback is not the end of the world.