Summer of generative AI: As it moves fast, wider appeal and availability is key - Hindustan Times
close_game
close_game

Summer of generative AI: As it moves fast, wider appeal and availability is key

Aug 28, 2023 06:42 PM IST

Despite companies, such as Google and Meta making rapid strides in AI tech, concerns about availability, regulation and ethical considerations continue to arise

We’re almost at the end of what’s definitely been the summer of artificial intelligence, or AI. One that has in more than definite terms, redefined the path of evolution for artificial intelligence, and specifically generative AI. The scope has widened, capabilities enhanced, with more relevance. If you thought the very realistic text-to-image tools were the pinnacle for generative AI, you have more to consider.

FILE PHOTO: Artificial Intelligence words are seen in this illustration taken March 31, 2023. REUTERS/Dado Ruvic/Illustration./File Photo(REUTERS) PREMIUM
FILE PHOTO: Artificial Intelligence words are seen in this illustration taken March 31, 2023. REUTERS/Dado Ruvic/Illustration./File Photo(REUTERS)

Sketches can now be turned into realistic images. That’s what Stable Doodle, Stability AI’s latest tool, attempts to do. It is available on the web and via Clipdrop on Android as well as iPhone. There’s a free version with limited functionality whereas a Pro version for about 350 per month that unlocks more options such as high-resolution background removal and access to Stable Diffusion XL generative AI.

This development effectively illustrates a sense of pace, at which the AI landscape is evolving. A blur of activity.

It is a charge spearheaded by generative AI, which finds more space in our lives and workflows. Chatbots, text-to-image systems, intelligent assistants, automated code generators, and AI-powered search, are some elements of new tools at our disposal. This summer, as Google, Microsoft and Apple focused extensively on AI albeit with differing approaches, it was a tip of the proverbial iceberg.

Elon Musk’s new xAI company is trying to “understand the true nature of the universe”. Google had made an AI-powered notes app called NotebookLM. Microsoft is testing an AI hub for Windows 11’s application store. Meta could release its LLaMA AI large language model (LLM) commercially.

But what does this mean for the often-bewildered consumers?

“Artificial Intelligence has the potential to revolutionize various aspects of our lives, and in many ways, it already has. AI algorithms help us in everything from simplifying mundane tasks to making life-changing decisions,” Sonam Srivastava, founder at investment advisory firm Wright Research, tells HT. She points to AI creating everything from data-driven investment portfolios to enabling faster diagnosis and treatment plans in healthcare.

Globalisation is a developing theme for chatbots and AI tools, in search of a wider user base. Google’s latest update for Bard, among functionality such as the ability to speak replies and export code, also adds support for 40 languages globally, with Indian language support including Hindi, Tamil, Telugu, Bengali, Kannada, Malayalam, Marathi, Gujarati and Urdu.

“It’ll be in line with our bold and responsible approach to AI. We’ve proactively worked with a number of policymakers, regulators and experts on this roll-out, and as we continue to evolve, we’ll keep our AI principles as a guide,” says Amar Subramanya, VP of Engineering at Google.

Adobe’s generative AI model Firefly, now supports 100 international languages, including Gujarati, Hindi, Malayalam, Marathi, Nepali, Punjabi, Tamil and Telugu. “We’ve been amazed at how creators have been using Firefly to create more than a billion gorgeous images and text effects making it one of Adobe’s most successful betas ever in just over three months,” says Ely Greenfield, CTO, Digital Media at Adobe.

For Adobe, this strengthening of language support further enhances the utility of its incredibly popular Photoshop app, and indeed Firefly’s standalone web tool, available to more users. Greenfield adds this “is about making Firefly accessible to more people in their preferred languages.” Firefly, launched in March, has since been integrated across Photoshop, Express and Illustrator apps, to name a few.

“Regular feedback loops involving human experts can also improve the performance and accuracy of AI models. It is also accompanied by ethical considerations and organizations need to establish guidelines and standards for responsible AI development and deployment,” Amit Choudhary, co-founder and CEO of skilling platform Saasguru, tells HT.

Microsoft is on course to integrate an AI assistant in Windows 11, called Copilot, with roll-out expected later this year. They say Copilot will summarise content you may be reading on any app. It’ll also see integration within Microsoft 365 apps, including Word.

The underlying approach was subtly tweaked too. Microsoft, by adopting the same open standard that OpenAI uses for ChatGPT plugins, hopes developers can now use one platform to build plugins that work across ChatGPT, Bing, and Microsoft 365 Copilot.

“This is one of the most powerful things for your developers, for every developer to be able to write a plugin and have it reach billions of users across all of these surface areas,” said Satya Nadella, executive chairman and CEO of Microsoft, at the BUILD keynote. Microsoft claimed a million active users for Bing within a month of its release, in February. There hasn't been a number refresh since. Copilot may invoke familiarity for many Windows 11 users.

A wider ecosystem of apps is helping Bard too, such as integration within Google Lens. In a follow-up to the text-to-image feature, Bard’s reverse search is now enabled. A user will be able to input an image for Bard to search with. Generative AI will respond with details about what it decodes.

We can trust Elon Musk to ask tough, existential questions. His new venture, xAI, doesn’t illustrate much in terms of what they intend to do for now. “Announcing the formation of @xAI to understand reality,” said Musk in a post on X (formerly Twitter), while the landing page on the company’s website, echoes similar thoughts. There are 12 researchers on board, an all-male team. Quite how this ‘understanding’ AI will evolve in the months to come, is anyone’s guess. Musk has often criticised OpenAI for close ties with Microsoft, which has invested heavily in the AI start-up and uses its GPT model for the Bing chatbot.

Some still have catching up to do. Meta, which spent a lot of its resources last year on the metaverse, is a commercial release of its LLaMA open-source LLM, to generate text, images and code. It’ll compete with OpenAI GPT-4 and PaLM2, Google’s LLM for Bard.

Avishek Gupta, CEO of investment advisory firm Caspian Debt, adopts a cautionary tone. “While AI has immense potential for positive impact, haphazard application can lead to significant detrimental impact. We are still very much in another hype cycle of AI,” he says.

Hints of less-than-optimal implementation are becoming clear.

The US Federal Trade Commission (FTC) is investigating OpenAI, over possible consumer harm. There are questions about how it vets' information used in training AI models, and the steps it takes to prevent false information from being shown in search results. When users log in to ChatGPT, a warning awaits – “The system may occasionally generate incorrect or misleading information and produce offensive or biased content.” There is the sense that OpenAI is leaving the risks to users to decode and isn’t doing enough to iron out generative AI’s deficiencies.

“To address inaccuracies, we need to invest in collecting high-quality, diverse data, and utilize more robust model validation techniques,” says Wright Research’s Srivastava. It is one of the reasons why talk of regulating AI, has gained traction.

This month, US lawmakers reintroduced the Algorithmic Justice and Online Platform Transparency Act, hoping for consensus that’ll make it mandatory for online platforms to disclose how their algorithms work, something that tech companies, social media sites and online shopping websites, keep behind closed doors.

According to the latest estimates by Precedence Research, the generative AI market will be worth $118.06 billion globally, by the year 2032. This is up from an estimated $13.71 billion in 2023. The need for regulation, amidst such a trajectory, gains paramount importance.

Earlier this summer, United Kingdom (UK) Prime Minister Rishi Sunak, while inaugurating the tenth edition of the London Tech Week, said the UK will take a global lead in regulating AI, with £100 million earmarked for a task force to work on AI standards and safeguards. “AI doesn’t respect traditional national borders. So, we need global cooperation between nations and labs,” Sunak said.

“Continual research and development are needed to enhance model interpretability, dataset quality, and algorithms,” says Mayank Singh, Co-Founder of Campus 365, an ed-tech platform. “Regulatory frameworks that promote justice and accountability as well as transparency in AI decision-making processes are essential, he adds.

There's more to come.

“AI's impact has been profound and pervasive. But we're just at the dawn of this technological era. As we delve deeper into the AI frontier, its role will only become more intrinsic to our everyday lives,” says Samanyou Garg, founder and CEO of Writesonic, an AI writing solutions platform.

An AI hub is coming soon with Windows 11 on millions of PCs globally, something already in the midst of extensive testing with preview builds ahead of a wider release. A number of closed group users are testing Google’s AI notebook, which attempts to create a personally trained AI – it’ll learn from data, documents and notes you point it to. For text-to-image tools, it is a big step that OpenAI’s Dall-E will train using Shutterstock’s massive library of images, videos and metadata, for six more years.

Unlock the power of data-driven insights with IIT Delhi's Data Science & Machine Learning Certificate Program! Click here to know more.

See more

Continue reading with HT Premium Subscription

Daily E Paper I Premium Articles I Brunch E Magazine I Daily Infographics
freemium
SHARE THIS ARTICLE ON
Share this article
SHARE
Story Saved
Live Score
OPEN APP
Saved Articles
Following
My Reads
Sign out
New Delhi 0C
Friday, June 21, 2024
Start 14 Days Free Trial Subscribe Now
Follow Us On