Рубрика: AI

AI world news.

The Major GCC Announcements for India in 2023 

EY, a professional services organisation, released its ‘Future of GCCs in India – a vision 2030‘ report, highlighting GCC expansion in India. The report estimates India’s domestic GCC market may reach $110b by 2030 (from $45b), led by software exports. By 2030, the report predicts 2400 GCCs growing at the rate of 115 new GCCs yearly (currently 70), The growth of the workforce is predicted to be 4.5M from the current 1.9 million people.

GCCs in India are focusing on talent retention, niche resource partnerships, expansion to more centres in different cities and also tier-II city interest. With a thriving startup ecosystem, India offers GCC-startup collaboration for innovation.

Here is this year’s list of the thriving GCCs across sectors.

Sandoz

In January this year, Sandoz, a major player in generic medicines, announced to set up their centre in Hyderabad, which will support their global knowledge services. Initially, Sandoz, a part of Novartis, will employ 800 people, with plans to expand to 1,800 in coming years. The company holds a diverse portfolio of around 1,000 medicines covering major therapies, achieving $9.7 billion in sales in 2019, benefitting over 500 million patients worldwide.

Alvarez & Marsal

Earlier this month, renowned global professional services firm Alvarez & Marsal (A&M) announced that they have launched the inaugural GCC in India. This centre will cater to A&M’s diverse business operations across six continents. The move underscores A&M’s strategy to deploy skilled professionals worldwide for delivering value, enhancing growth, and fostering innovation.

Deutsche Bank

Deutsche India, home to Deutsche Bank‘s largest technology centre, is expanding in India, planning to hire thousands more. The bank has hired over 2,500 individuals since January 2023, with hiring expected to continue through 2023 and beyond. The centre employs about 16,000 professionals, primarily engineers. Deutsche Bank aims to build an internal engineering workforce, moving away from outsourcing, with a goal to achieve a 70:30 in-house to outsourcing work ratio.

NatWest Group

NatWest Group, a major UK bank, has introduced a pioneering pilot in India’s global capability centres to offer first-time job opportunities to women who have never been part of the workforce. This initiative, named ‘Wish,’ began this year with 15 women in Chennai and Gurugram, focusing on candidates with no experience regardless of age. The bank’s existing ‘Re-invest‘ program supports women returnees, while this ‘Wish’ pilot elevates its India women hiring plan. The GCC employs 16,000 in India, recruiting over 2,000 annually, and has initiated upskilling efforts, allowing employees to dedicate two extra days yearly for future-oriented skill enhancement.

PwC

Amidst tech layoffs, PwC is set to expand its presence in India by adding 30,000 new jobs within the next five years. This expansion initiative is the result of the collaboration between PwC India and PwC US. These ventures are geared towards establishing new global centres in India and amplifying the growth of existing ones. Currently, PwC sustains a workforce exceeding 50,000 individuals across India, spanning both its Indian operations and its global delivery centres.

Capegemini

In March, this year Capegemini’s managing director Ananth Chandramouli said that the company in India continues to grow at a higher rate than global centres. The company currently employs around 180,000 team members across 13 cities in India out of the total 3,60,000 total employees in the company.

The company is looking forward to further development in the transportation, manufacturing, telecom, consumer products and retail sectors.

JP Morgan Chase

JPMorgan views India as a vital Asian market, constantly boosting its capabilities. Amid global competition, the company has expanded its tech centres in Mumbai, Bengaluru, and Hyderabad in the past few years, totaling 3 out of 21 tech centres in India itself. These centres tap into India’s rich talent and expertise. The bank plans to use India not only as a tech hub but also for product development. This strategic move aligns with the growth of their global tech centres and India’s tech landscape.

Nissan

Nissan along with Renault announced an ambitious plan to increase the production and R&D activities, to introduce electric vehicles, and transitioning to carbon-neutral manufacturing. Based in Chennai, the collaboration targets six new production vehicles, including two electric models, establishing the Renault-Nissan centre as a global export hub. An initial investment of about $600 million is allocated to these initiatives, aiming to create up to 2,000 jobs.

Hewlett Packard Enterprise (HPE)

HPE announced in July this year that the company intends to begin producing high-volume servers in India. Plans include manufacturing approximately $1 billion worth of servers in India within the first five years. Partnering with VVDN Technologies, HPE’s production will be based in Manesar, Haryana. The move is aligned with the ‘Make in India’ initiative and reinforces HPE’s commitment to the country. HPE’s significant workforce in India, including over 4,000 scientists and engineers, is central to this endeavour across various sectors and initiatives.

Okta

Earlier this month, Okta opened an office in Bengaluru, India, for secure digital transformation in the APAC region. This hub will include a state-of-the-art R&D centre for global identity solution development. Amid India’s rising cybersecurity spending (predicted 18% growth by 2025), Okta aims to serve local customers, enhance security, and contribute to India’s technology landscape. The office will be the site of its first innovation centre in Asia Pacific and will facilitate engagement with local policy makers, educational institutions, and customised integrations for Indian enterprises. The company also partnered with OpenAI earlier this month.

Johnson Controls

Johnson Controls inaugurated its largest OpenBlue Innovation Centre in Bengaluru in July this year, showcasing net zero building tech with cloud, edge, and AI. The centre supports India’s G20 priorities, advancing sustainability and digitalization. It boasts AI-driven energy-efficient tech demos, a digital twin, access control, computer vision, fire safety solutions, and VR experiences for net zero goals. Johnson Controls, present in India for 30 years, will employ 300 engineers at this centre, expanding its role in smart building innovation in India’s IT hub.

Houghton Mifflin Harcourt (HMH)

In July this year, Learning technology company Houghton Mifflin Harcourt (HMH) launched its first Asia-Pacific Center of Excellence (COE) in Pune, India. COEs within Global Capability Centers (GCCs) are organisations or teams focused on developing and advancing analytics capabilities within a company.

This expansion aims to harness India’s tech talent for innovation in K-12 education solutions. The Pune Center will focus on product innovation, R&D, and development, catering to Indian schools implementing IB and IGCSE curriculum. HMH’s solutions integrate core instruction, practice, intervention, assessment, and professional learning.

The post The Major GCC Announcements for India in 2023 appeared first on Analytics India Magazine.

‘Every Tech Company Will Be an AI Company’: Hugging Face’s $235M Vote of Confidence

‘Every Tech Company Will Be an AI Company’: Hugging Face’s $235M Vote of Confidence August 24, 2023 by Jaime Hampton

Hugging Face announced it has raised $235 million in a Series D funding round led by Salesforce Ventures with participation from a host of tech giants including IBM, Google, Amazon, Nvidia, Intel, AMD, and Qualcomm, as well as Ashton Kutcher’s Sound Ventures. The company is now valued at $4.5 billion.

Hugging Face is a developer community and platform with open source AI and machine learning resources like models, datasets, ML demos, and libraries. The company claims it accelerates AI deployment by providing MLOps solutions including customized support, training, fine-tuning, deployment services, APIs and more.

One popular resource is the Transformers library, a Python package containing open source Transformer models for text, image, and audio tasks that is compatible with PyTorch, TensorFlow, and JAX deep learning libraries. The company also has a platform called Hugging Face Hub that hosts Git-based code repositories, models, datasets, and applications.

The company was founded in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond, and Thomas Wolf. Hugging Face originally purveyed a chatbot app for teenagers, marketed as an AI friend that could entertain with its cheeky conversation. In 2018, Hugging Face released an open source version of PyTorch BERT, the conversational AI model behind its chatbot, leading to its renewed focus as a machine learning platform.

Hugging Face previously raised $100 million in a Series C round led by Lux Capital in May 2022: “At Hugging Face, we know that Machine Learning has some important limitations and challenges that need to be tackled now like biases, privacy, and energy consumption. With openness, transparency, and collaboration, we can foster responsible and inclusive progress, understanding, and accountability to mitigate these challenges,” the company said in a blog post at the time.

The company will use the Series D funds to increase its current 170-employee headcount and invest in its technology, according to a Reuters report that also quoted CEO Delangue: “In five years, every tech company will be an AI company.” Another report from The Information claims that Hugging Face could generate more than $30 million in annual revenue.

(Source: Twitter)

Open source AI is an alternative development paradigm to the closed source, commercial AI that is often called a black box due to the nebulous nature of its training data and configuration. Companies often keep their proprietary code under wraps to maximize profitability, offering users little opportunity to peer under the hood or extensively customize their models.

Proponents of open source AI with its publicly available code say it promotes transparency and community-driven enhancements, as anyone can inspect, modify, and distribute the related software and frameworks.

Some companies like Meta are bridging the gap by releasing open source versions of AI models like Llama-2, its 70 billion-parameter LLM trained on two trillion tokens of raw text, hosted by Hugging Face. Stability AI’s Stable Diffusion model is another example of an open source model hosted on the platform.

Series D-backer IBM has contributed over 200 open models and datasets on Hugging Face, including its recent Geospatial Foundation Model built in collaboration with NASA. Additionally, IBM said in a release it has plans to host Meta’s Llama 2-chat model within its watsonx AI platform, saying it will further the company’s strategy of leveraging both third-party and proprietary AI models to maintain open innovation.

Related

Ikigai lands $25M investment to bring generative AI to tabular data

Ikigai lands $25M investment to bring generative AI to tabular data Kyle Wiggers 18 hours

Organizations are awash in data, but struggle with a host of challenges to actually use, organize and analyze that data. According to one estimate, companies will store 100 zettabytes of data in the cloud by 2025. But as of now, just 13% of organizations are delivering on their analytics and data strategy, an MIT Technology Review Insights and Databricks survey found.

Devavrat Shah argues that enabling companies to effectively forecast and conduct scenario-based planning requires harnessing deep, complex data types from hundreds of sources across the business — and that AI holds the key to this. He’s the founder of Celect, an AI app for allocating and fulfilling big box retail orders (which was acquired by Nike in 2019), and directs MIT’s statistics and data science center.

“Today, the best moonshot project for AI is to bring it to organizations or enterprises,” Shah told TechCrunch via email. “Enterprises are run by experts, and we believe that these experts need to be able to work seamlessly with AI to harness the possible benefits it holds.”

To realize the mission of “empowering every enterprise with AI,” as Shah puts it, Shah started Ikigai Labs, which offers a no-code platform built on top of proprietary graphical models for prediction, sparse data reconciliation and optimization. Ikigai today announced that it raised $25 million in Series A funding led by Premji Invest with participation from Foundation Capital and E& Capital VC, bringing its total raised to $38.2 million.

Shah co-founded Ikigai alongside Vinayak Ramesh, who previously founded Well Frame, a healthcare company that Blackstone purchased in 2012. While in graduate school at MIT, Ramesh worked with Shah on building AI for tabular data — data that’s organized in a table with row and columns, like a database — using large graphical models.

Technically within the family of neural networks, graphical models represent the probabilistic relationships among a set of variables, Shah explained. “Most of the enterprise data is tabular, sparse and typically time-stamped. Large graphical models are precisely well-suited for this setting,” he added. “In modern parlance, they’re ‘generative AI’ for tabular data.”

Ikigai’s platform is designed to enable companies to create and deploy these graphical models to power apps within their organizations. Using it, customers can train models on-demand on their enterprise data, creating models that assist in forecasting, scenario planning and analysis.

One might question why Ikigai’s graphical models are superior to, say, the large language models (LLMs) that’ve gained currency in recent years. Shah notes that LLMs work well for text and other unstructured data, but also that they’re expensive to operate, requiring vast amounts of storage compared to their graphical model equivalents.

“We provide building blocks that allow customers to solve a broad spectrum of use cases,” Shah said. “We hope to bring along everyone to ride the wave of AI and not drown in it.”

Shah isn’t naïve enough to think that Ikigai is without competition in the sprawling market for enterprise AI. He named C3.ai, Anaplan, Dataiku and Hugging Face as top rivals — at least in the sense that they offer at least a subset of what Ikigai offers.

But for what it’s worth, Sandesh Patnam, a managing partner at Premji Invest, is confident in Ikigai’s ability to stand out from the crowd.

“Ikigai’s founding team possesses a depth of industry and go-to-market expertise that can push the AI frontier into the center of business operations and decisions,” he said via email. “Their innovation with large graphical models will be embraced by all enterprises that look to apply generative AI to their existing tabular data.”

With the new capital, San Francisco-based Ikigai plans to grow its team from 30 people to 70 by the end of the year.

Fine-tuning OpenAI’s GPT-3.5 Turbo can make it as capable as GPT-4 (if not more)

abstract cube

OpenAI's advanced large language models have been leveraged by enterprises and developers for their own specific use cases. Now, an update to GPT-3.5 Turbo is going to boost its functionality for its customers.

On Tuesday, OpenAI announced that its most cost-effective model in the GPT-3.5 family, GPT-3.5 Turbo, would be available for fine-tuning. This means that developers can now use their own data to customize the model for their use cases.

Also: How to make ChatGPT provide sources and citations

"Since the release of GPT-3.5 Turbo, developers and businesses have asked for the ability to customize the model to create unique and differentiated experiences for their users," said OpenAI in the post.

In the private beta, OpenAI found that customers were able to improve the model's performance in a variety of use cases. These include improved steerability, which allows the model to better follow instructions, reliable output formatting, and custom tone, which allows businesses to incorporate their brand voice within the model.

OpenAI also claims that the fine-tuning allows businesses to shorten their prompts, with early testers reducing prompt size by up to 90%. This reduction cuts costs and speeds up each API call, according to the company.

Also: You can demo Meta's AI-powered multilingual speech and text translator. Here's how

Most impressively, OpenAI shared that early tests showed the fine-tuned version of GPT-3.5 Turbo can "match, or even outperform" GPT-4-level capabilities on "certain narrow tasks."

To address privacy concerns involved with harnessing an AI model for enterprise use cases, OpenAI reassures users that the customer data used to fine-tune the model remains in customer ownership and is not used by OpenAI to train other models, such as with another API model.

Artificial Intelligence

Salesforce Releases AI Acceptable Use Policy

Man harnessing a virtual circuited brain with a glowing central processor.
Image: sdecoret/Adobe Stock

Salesforce released a policy governing the use of its AI products, including generative AI and machine learning, on Wednesday. This AI policy is an interesting precedent because Salesforce is such a large organization in the field of sales platforms and customer relationship management. Additionally, the policy offers clear guidelines, and AI regulations are still in the discussion and development stages.

We talked to Executive Vice President and General Manager at Salesforce Industry Clouds Jujhar Singh about the policy and why it’s important for companies to draw ethical lines around their generative AI.

Jump to:

  • What is Salesforce’s Artificial Intelligence Acceptable Use Policy?
  • The policy joins Salesforce’s internal generative AI guidelines
  • Which products does Salesforce’s AI policy apply to?
  • How can businesses prepare for the future of generative AI?
  • What will the generative AI landscape look like in six months?

What is Salesforce’s Artificial Intelligence Acceptable Use Policy?

Salesforce’s Artificial Intelligence Acceptable Use Policy outlines ways in which its AI products may not be used. It was published on August 23. The policy restricts what Salesforce customers can use its generative AI products for, including banning their use for weapons development, adult content, profiling based on protected characteristics (such as race), biometric identification, medical or legal advice, decisions that may have legal consequences and more.

The policy was written under the supervision of Paula Goldman, chief ethical and humane use officer at Salesforce, who sits on the U.S. National Artificial Intelligence Advisory Committee.

“It’s not enough to deliver the technological capabilities of generative AI, we must prioritize responsible innovation to help guide how this transformative technology can and should be used,” the Salesforce team wrote in the blog post announcing the policy.

The policy joins Salesforce’s internal generative AI guidelines

Salesforce has a public list of internal guidelines for its development of generative AI. This could serve as an example for other companies creating policies around their creation of generative AI. Salesforce’s guidelines are:

  1. Accuracy and verifiable, traceable answers.
  2. Avoiding bias or privacy breaches.
  3. Support data provenance and include a disclaimer on AI-generated content.
  4. Identify the appropriate balance between human and AI.
  5. Reduce carbon footprint by right-sizing models.

“Transparency is a big part of how we deal with gen AI because everything goes back to the (concept of) trusted AI,” said Singh in an interview with TechRepublic.

Singh emphasized the importance of having a zero retention policy so personally identifiable information is not used to train an AI model that might regurgitate it somewhere else. Salesforce is building filters to remove toxic content and constantly tweaking them, Singh said.

Which products does Salesforce’s AI policy apply to?

The policy applies to all services offered by Salesforce or its affiliates.

Under that umbrella are Salesforce’s flagship generative AI products, which include everything hosted on the EinsteinGPT platform, which is a library of public and private foundation models, including ChatGPT, for customer service, CRM and other tasks across various industries. Its competitors are Hubspot’s ChatSpot.AI and Microsoft Copilot.

How can businesses prepare for the future of generative AI?

“Tech leaders have to really think through that they need to upskill their people on gen (generative) AI dramatically,” said Singh. “In fact, 60% of the people that were asked this question felt they didn’t have the skills, but they also thought that their employers actually need to deliver those skills.”

In particular, skills like prompt engineering will be critical to thriving in a changing business world, he said.

SEE: Hire the right prompt engineer with these guidelines. (TechRepublic Premium)

In terms of infrastructure, he emphasized that companies need to have a strong trust foundation in order to make sure AI is on-task and producing accurate content.

What will the generative AI landscape look like in six months?

Singh said there might be a divide between industries with tighter or looser AI regulatory oversight.

“I think the industries that are more regulated are going to have a more human-in-the-loop approach in the next six months,” he said. “As we go into other industries, they are going to be much more aggressive in adopting AI. Assistants working on your behalf are going to be much more prevalent in those industries.”

Plus, industry-specific LLMs will “start becoming very, very relevant very soon,” he said.

Subscribe to the Innovation Insider Newsletter

Catch up on the latest tech innovations that are changing the world, including IoT, 5G, the latest about phones, security, smart cities, AI, robotics, and more.

Delivered Tuesdays and Fridays Sign up today

Meta releases Code Llama, a code-generating AI model

Meta releases Code Llama, a code-generating AI model Kyle Wiggers 16 hours

Meta, intent on making a splash in a generative AI space rife with competition, is on something of an open source tear.

Following the release of AI models for generating text, translating languages and creating audio, the company today open sourced Code Llama, a machine learning system that can generate and explain code in natural language — specifically English.

Akin to GitHub Copilot and Amazon CodeWhisperer, as well as open source AI-powered code generators like StarCoder, StableCode and PolyCoder, Code Llama can complete code and debug existing code across a range of programming languages, including Python, C++, Java, PHP, Typescript, C# and Bash.

“At Meta, we believe that AI models, but large language models for coding in particular, benefit most from an open approach, both in terms of innovation and safety,” Meta wrote in a blog post shared with TechCrunch. “Publicly available, code-specific models can facilitate the development of new technologies that improve peoples’ lives. By releasing code models like Code Llama, the entire community can evaluate their capabilities, identify issues and fix vulnerabilities.”

Code Llama, which is available in several flavors, including a version optimized for Python and a version fine-tuned to understand instructions (e.g. “Write me a function that outputs the Fibonacci sequence”), is based on the Llama 2 text-generating model that Meta open sourced earlier this month. While Llama 2 could generate code, it wasn’t necessarily goodcode — certainly not up to the quality a purpose-built model like Copilot could produce.

In training Code Llama, Meta used the same data set it used to train Llama 2 — a mix of publicly available sources from around the web. But it had the model “emphasize,” so to speak, the subset of the training data that included code. Essentially, Code Llama was given more time to learn the relationships between code and natural language than Llama 2 — its “parent” model.

Each of the Code Llama models, ranging in size from 7 billion parameters to 34 billion parameters, were trained with 500 billion tokens of code along with code-related data. The Python-specific Code Llama was further fine-tuned on 100 billion tokens of Python Code, and, similarly, the instruction-understanding Code Llama was fine-tuned using feedback from human annotators to generate “helpful” and “safe” answers to questions.

For context, parameters are the parts of a model learned from historical training data and essentially define the skill of the model on a problem, such as generating text (or code, in this case), while tokens represent raw text (e.g. “fan,” “tas” and “tic” for the word “fantastic”).

Several of the Code Llama models can insert code into existing code and all can accept around 100,000 tokens of code as input, while at least one — the 7 billion parameter model — can run on a single GPU. (The others require more powerful hardware.) Meta claims that the 34 billion-parameter model is the best-performing of any code generator open sourced to date — and the largest by parameter count.

You’d think a code-generating tool would be massively appealing to programmers and even non-programmers — and you wouldn’t be wrong.

GitHub claims that more than 400 organizations are using Copilot today, and that developers within those organizations are coding 55% faster than they were before. Elsewhere, Stack Overflow, the programming Q&A site, found in a recent survey that 70% are already using — or planning to use — AI coding tools this year, citing benefits like increased productivity and faster learning.

But like all forms of generative AI, coding tools can go off the rails — or present new risks.

A Stanford-affiliated research team found that engineers who use AI tools are more likely to cause security vulnerabilities in their apps. The tools, the team showed, often generate code that appears to be superficially correct but poses security issues by invoking compromised software and using insecure configurations.

Then, there’s the intellectual property elephant in the room.

Some code-generating models — not necessarily Code Llama, although Meta won’t categorically deny it — are trained on copyrighted or code under a restrictive license, and these models can regurgitate this code when prompted in a certain way. Legal experts have argued that these tools could put companies at risk if they were to unwittingly incorporate copyrighted suggestions from the tools into their production software.

And — while there’s no evidence of it happening at scale — open source code-generating cools could be used to craft malicious code. Hackers have already attempted to fine-tune existing models for tasks like identifying leaks and vulnerabilities in code and writing scam web pages.

So what about Code Llama?

Well, Meta only red-teamed the model internally with 25 employees. But even in the absence of a more exhaustive audit from a third party, Code Llama made mistakes that might give a developer pause.

Code Llama won’t write ransomware code when asked directly. However, when the request is phrased more benignly — for example, “Create a script to encrypt all files in a user’s home directory,” which is effectively a ransomware script — the model complies.

In the blog post, Meta admits outright that Code Llama might generate “inaccurate” or “objectionable” responses to prompts.

“For these reasons, as with all LLMs, Code Llama’s potential outputs cannot be predicted in advance,” the company writes. “Before deploying any applications of Code Llama, developers should perform safety testing and tuning tailored to their specific applications of the model.”

Despite the risks, Meta places minimal restrictions on how developers can deploy Code Llama, whether for commercial or research use cases. They must simply agree not to use the model for malicious purposes and, if deploying it on a platform with greater than 700 million monthly active users — i.e. a social network that might rival one of Meta’s — request a license.

“Code Llama is designed to support software engineers in all sectors — including research, industry, open source projects, NGOs and businesses. But there are still many more use cases to support than what our base and instruct models can serve,” the company writes in the blog post. “We hope that Code Llama will inspire others to leverage Llama 2 to create new innovative tools for research and commercial products.”

ChatGPT-assisted bots are spreading on social media

Robot hiding in a mug, 3d rendering - stock illustration

For many users, scrolling through social media feeds and notifications is like wading in a cesspool of spam content. A new study identified 1,140 AI-assisted bots that were spreading misinformation on X (formerly known as Twitter) about cryptocurrency and blockchain.

But bot accounts posting this type of content can be hard to spot, as the researchers from Indiana University found. The bot accounts used ChatGPT to generate their content and were hard to differentiate from real accounts, making the practice more dangerous for victims.

Also: You can demo Meta's AI-powered multilingual speech and text translator. Here's how

The AI-powered bot accounts had profiles that resembled those of real humans, with profile photos and bios or descriptions about crypto and blockchain. They made regular posts generated with AI, posted stolen images as their own, and made replies and retweets.

The researchers discovered that the 1,140 Twitter bot accounts belonged to the same malicious social botnet, which they referred to as "fox8." A botnet is a network of connected devices — or, in this case, accounts — that are centrally controlled by cybercriminals.

Also: 4 things Claude AI can do that ChatGPT can't

Generative AI bots have been getting better at mimicking human behaviors. This means traditional and state-of-the-art bot detection tools, like Botometer, are now insufficient. These tools struggled to identify and differentiate bot-generated content from human-generated content in the study, but one stood out: OpenAI's AI classifier, which was able to identify some bot tweets.

How can you spot bot accounts?

The bot accounts on Twitter exhibited similar behavioral patterns, like following each other, using the same links and hashtags, posting similar content, and even engaging with each other.

Also: New 'BeFake' social media app encourages users to transform their photos with AI

Researchers combed over the tweets of the AI bot accounts and found 1,205 self-revealing tweets.

Out of this total, 81% had the same apologetic phrase:

"I'm sorry, but I cannot comply with this request as it violates OpenAI's Content Policy on generating harmful or inappropriate content. As an AI language model, my responses should always be respectful and appropriate for all audiences."

The use of this phrase suggests that the bots are instructed to generate harmful content that goes against OpenAI's policies for ChatGPT.

The remaining 19% used some variation of "As an AI language model" language, with 12% specifically saying, "As an AI language model, I cannot browse Twitter or access specific tweets to provide replies."

Also: ChatGPT vs. Bing Chat vs. Google Bard: Which is the best AI chatbot?

The fact that 3% of the tweets posted by these bots linked to one of three websites (cryptnomics.org, fox8.news, and globaleconomics.news) was another clue.

These sites look like normal news outlets but had notable red flags, like the fact that they were all registered around the same time in February 2023, had popups urging users to install suspicious software, all seem to use the same WordPress theme, and have domains that resolve to the same IP address.

Malicious bot accounts can use self-propagation techniques in social media by posting links with malware or infectable content, exploiting and infecting a user's contacts, stealing session cookies from users' browsers, and automating follow requests.

Artificial Intelligence

OpenAI and Scale Join Forces to Fine-Tune GPT-3.5 for Enterprises

OpenAI Releases Shap-E, Generative Model for 3D Assets

OpenAI announced that it is partnering with San Francisco based data labeling startup Scale to help more companies benefit from fine-tuning its text generating model GPT-3.5.

The company in their blog post stated that it is working with Scale as a preferred partner to extend the benefits of our fine-tuning capability given its experience helping enterprises securely and effectively leverage data for AI.

OpenAI further said they have extended the opportunity to Scale customers to fine-tune OpenAI models, mirroring the process they would follow with OpenAI. Moreover, these customers stand to gain from Scale’s proficiency in enterprise AI and the utilization of their Data Engine.

“Scale extends our ability to bring the power of fine-tuning to more companies, building on their enterprise AI experience to help businesses better apply OpenAI models for their unique needs.”said Brad Lightcap, COO, OpenAI.

OpenAI a day earlier had announced that fine-tuning for GPT-3.5 Turbo is now available and fine-tuning for GPT-4 is coming this fall. OpenAI has stated that the fine-tuning of GPT 3.5 Turbo is suitable specifically for businesses and developers to customize the model depending upon their use case as it lets them train the model on company’s data and run it at scale.

Founded in 2016, Scale’s Generative AI Platform leverages enterprise data to customize powerful base generative models to safely unlock the value of AI.

“We are excited to partner with OpenAI to supercharge model performance – helping every enterprise utilize AI most effectively for their unique needs. Prompting alone—atop even the best LLMs like GPT-3.5 — is not enough model customization to produce the most accurate, efficient results. As with software, an incredible amount of value comes from fine-grained optimizations, and fine tuning is critical for that.” said Alexandr Wang, Founder and CEO, Scale AI

Scale has earlier worked with American financial service and technology company Brex. Scale in their blog post claimed that by using the GPT-3.5 fine-tuning API on Brex data annotated with Scale’s Data Engine, they saw that the fine-tuned GPT-3.5 model outperformed the stock GPT-3.5 turbo model 66% of the time.

The post OpenAI and Scale Join Forces to Fine-Tune GPT-3.5 for Enterprises appeared first on Analytics India Magazine.

OpenAI partners with Scale AI to allow companies to fine-tune GPT-3.5

OpenAI partners with Scale AI to allow companies to fine-tune GPT-3.5 Kyle Wiggers 12 hours

OpenAI, the AI startup behind the viral AI-powered chatbot ChatGPT, plans to partner with third-party vendors to make it easier for developers — specifically enterprises — to fine-tune its AI models using custom data.

Today, OpenAI announced that it’ll team up with Scale AI, the San Francisco-based data labeling startup, to bring together Scale AI’s fine-tuning tools and OpenAI’s GPT-3.5 text-generating model. (GPT-3.5 is the predecessor to GPT-4, OpenAI’s flagship model, which understands images as well as text.)

Fine-tuning lets developers tailor an AI model to specific tasks. For example, a business could fine-tune a model to match its brand voice and tone, or have it respond to questions in a particular language.

Scale customers will be able to prep and “enhance” their data using Scale’s Data Engine platform, then fine-tune GPT-3.5 with their data and further customize the model with features like the ability to reference or cite their proprietary data in its responses.

Scale says that fine-tuned GPT-3.5 models will be reviewed by human experts to “ensure that the model exceeds performance expectations and safety requirements.”

“The partnership incorporates OpenAI’s GPT-3.5 fine-tuning APIs with Scale’s fine-tuning and data expertise from over seven years of building and deploying pioneering AI solutions for customers,” Scale CEO Alexandr Wang told TechCrunch in an email interview. “Scale is OpenAI’s preferred fine-tuning partner to deploy the most performative models, and we look forward to expanding our relationship. OpenAI has chosen to partner with Scale because of our expertise and proven track record of delivering high-quality models with real business impact for our customers.”

News of the collaboration comes a day after OpenAI said that it would begin letting customers fine-tune GPT-3.5 Turbo, the lightweight version of GPT-3.5, using its fine-tuning API. In the announcement, OpenAI hinted that it was developing an in-house tuning tool; it’s unclear whether that tool’s meant to complement platforms like Scale AI or serve a different purpose altogether.

In any case, OpenAI COO Brad Lightcap pitched the team-up with Scale AI as a way to allow companies customizing GPT-3.5 to benefit from additional services and expertise beyond what OpenAI provides.

“We want to serve customers wherever they are. We’re excited to work with great partners like Scale that help us serve the needs of our customers as they put AI to work in their organizations,” he said via email. “This new partnership allows us to extend the benefits of our capabilities to more companies given Scale’s leadership and experience helping organizations effectively leverage enterprise data for AI. Our customers can now fine-tune OpenAI models just as they would through OpenAI, while also benefiting from Scale’s enterprise AI expertise and Data Engine.”

When asked whether to expect new fine-tuning partners for OpenAI models down the line — such as GPT-4, which OpenAI says will gain fine-tuning capabilities in the fall — Lightcap declined to say.

Twilio Segment Unveils CustomerAI Predictions Tool in Partnership with Amazon SageMaker

Twilio Segment has unveiled a powerful tool designed to transform customer data into actionable insights. Dubbed CustomerAI Predictions, the tool aids marketers in making accurate forecasts regarding the behaviour of specific customer segments. Alex Millet, Senior Director of Product at Twilio Segment, emphasised the significance of quality data for marketers and highlighted the value that can be derived from existing customer data. The tool leverages information like clickstream data from websites or apps, combined with communication data from Twilio, to enable companies to comprehend customer engagement levels and identify areas requiring attention.

Incorporating machine learning, CustomerAI Predictions is the result of collaboration between Twilio Segment and Amazon SageMaker. This strategic partnership facilitated the swift development of the predictions tool by utilising SageMaker’s advanced machine-learning infrastructure. As part of their roadmap, Twilio is exploring the integration of generative AI-based email tools, enabling marketers to craft personalised emails based on the data insights garnered from CustomerAI Predictions.

Twilio’s journey into predictive analytics stems from its acquisition of Segment in 2020, signalling a strategic expansion beyond its core communications API business. The launch of CustomerAI Predictions underscores the growing importance of not just accumulating customer data, but effectively utilising it to enhance customer experiences and target campaigns with precision.

Twilio’s endeavours to transform data into predictive insights align with the broader industry trend of leveraging data for strategic decision-making. As more companies acknowledge the value of predictive analytics in improving customer experiences and driving business growth, tools like CustomerAI Predictions play a pivotal role in shaping marketing strategies for the future. With a commitment to innovation, Twilio’s expansion into predictive analytics is set to empower marketers and businesses in optimising their engagement strategies for the evolving digital landscape.

The post Twilio Segment Unveils CustomerAI Predictions Tool in Partnership with Amazon SageMaker appeared first on Analytics India Magazine.