AI model madness vs. AI model skeptics – SiliconANGLE

New artificial intelligence models keep arriving every day — make that several times a day now, judging from the list below — and you have to wonder if customers can keep up. What do you bet your company on, when an even shinier new one arrives by lunchtime?

Moreover, doubts remain about whether the current crop will be enough to get AI where all its fans hope. What all these models need to get better, among other things, is more data — real, not synthetic data, which can lead to model collapse.

Problem is, the people providing all that data — novelists, actors, musicians and, um, journalists — remain unhappy and litigious. The divide is only growing wider, as Perplexity’s perplexing post blaming media makes apparent. Maybe getting $500 million more in funding at an $8 billion valuation makes you feel a little bold.

Given the glut of models, it’s no surprise that a lot of AI services being rolled out from Microsoft, UiPath, Cisco Systems, Informatica and more are in the vein of making it easier to use the models. That includes agents, such as Anthropic’s new Claude model that lets AI models connect directly to your PC — though the security implications of that are scary.

Still, AI also isn’t yet lifting all enterprise boats, as earnings this week from IBM, ServiceNow, SAP and others signal a mixed bag for tech suppliers. We might get more answers from next week’s earningspalooza, which includes Alphabet/Google, AMD, Microsoft, Meta, Amazon, Apple and Intel.

In other news besides AI this week, Arm threatened to cancel Qualcomm’s license to use its instruction set, but it’s not yet clear if that’s just a negotiating tactic. The next 60 days should be interesting.

The platforming of cybersecurity — I can’t bring myself to use the “word” platformization — is prompting some serious alliances, such as IBM and Palo Alto Networks. Dave Vellante’s latest Breaking Analysis interview with execs from each company lays it all out.

And late-breaking Friday, self-driving car company Waymo announced a massive $5.6 billion funding round for its autonomous taxi service from parent Alphabet and others.

TheCUBE Research analysts John Furrier and Dave Vellante discuss this and other news in more detail on this week’s theCUBE Pod, already out on YouTube. And don’t miss Vellante’s weekly deep dive, Breaking Analysis, arriving this weekend for some lean-back reading.

Here’s what happened this week around the enterprise:

AI and data: New AI models and their skeptics

Headline news and trends

Andy Kessler is an AI skeptic, at least in the short term, in his Wall Street Journal column: AI can’t teach AI new tricks: A lack of human prose leads to what’s called ‘model collapse’ and pure gibberish.

Longer-term, which could still be relatively short in AI time, I wonder if there will be more of a focus beyond language as the chief AI driver. After all, humans learn to function fundamentally by moving through space and time, secondarily through language. That’s why we can learn and adapt to new situations so fast, with very few examples and data, and we don’t hallucinate like large language models (though perhaps that’s less obvious in this political season). I think we’re likely to see much more capable AIs that are trained on motion as well as language — call them large motion models? Some people already have. Or maybe small motion models. But I’m thinking even more broadly, like Numenta founder Jeff Hawkins, who contends the human brain actually uses tens of thousands of models. Which is to say, we have a long way to go before we get to true intelligence and reasoning by AIs, let alone artificial general intelligence. That doesn’t mean we don’t have enormous opportunities in current models, but let’s not fool ourselves that they’re the end game.

White House issues memorandum on AI use in national security initiatives

Thousands of creatives join forces to combat AI data scraping

This kind of adversarial attitude by Perplexity, not really backed by facts, could come back to bite it. Seems unwise. As TechCrunch puts it: “Throughout the blog, Perplexity does not mention or address the central claim of the lawsuit: that Perplexity allegedly copies content at a massive scale from publishers, then competes with them for the same audience.” Meantime, Meta tries to make nice: Meta inks multiyear AI content licensing deal with Reuters

U.K. watchdog probes Alphabet’s deal with Anthropic — not long after it started doing the same for Amazon’s Anthropic deal.

LinkedIn hit with $334M fine for violating GDPR

New models and services

IBM releases new Granite foundation models under ‘permissive’ Apache license

Microsoft brings autonomous agents to Dynamics 365 and Copilot Studio

UiPath is marrying generative AI with robots to accelerate business automation

And some analysis from UiPath Forward: AI agents for automation reshape enterprise efficiency and decision-making: Industry analysts weigh in and The next phase of AI: Orchestrating agents and robots to streamline enterprise workflows

Anthropic releases improved Claude models that can control your computer Rachel Tobac of SocialProof Security, however, worries about the security implications, tweeting in a thread worth reading in full: “Breaking out into a sweat thinking about how cyber criminals could use this tool. This easily automates the task of getting a machine to go to a website and download malware or provide secrets, which could scale attacks (more machines hacked in a shorter period of time).”

Anthropic introduces a new feature that lets it write data analysis with JavaScript

Stability AI releases next-gen open-source Stable Diffusion 3.5 text-to-image AI model family

Cohere announces Aya Expanse multilingual AI model family for researchers

Cisco’s Webex Contact Center platform gets agentic AI-powered chatbots And analysis from Zeus Kerravala: Analysis: At WebexOne, Cisco debuts AI-driven tools to enhance employee collaboration and customer service

Meta debuts slimmed-down Llama models for low-powered devices

Meta’s Spirit LM generates more expressive voices that reflect anger, surprise, happiness and other emotions

Haiper debuts new flagship video generation model

Runway’s Act-One uses smartphone cameras to replicate facial expression motion capture

JetBrains launches Mellum, an AI model built to assist developers with code completion

Genmo introduces Mochi 1, an open-source text-to-video generation model

Informatica’s blueprints outline how to build generative AI apps in the cloud

Asana’s AI Studio promises to transform work management with smart workflows and task automation

Freddy AI Agent from Freshworks promises quick deployment and autonomous support

Zencoder launches to bring AI coding agents to software development

Doubling down on AI and splitting at BMC Connect 2024

Well, that’s a niche application: Law enforcement tech startup Flock Safety uses AI to shut down muscle car sideshows

Money matters

Report: AI search startup Perplexity in talks with investors over $500M funding

Pegasystems tops Q3 earnings estimates, stock rises 12%

Thomson Reuters acquires agentic AI accounting assistant firm Materia

Carbon Robotics raises $70M to scale up AI-powered robotic farming solutions

Concentric bags $45M to expand its AI-powered data cataloging tools

Stream.Security secures $30M to enhance AI-powered cloud threat management

Fixify nabs $25M for its AI-supported help desk service

Pantheon AI, a San Francisco-based AI-enabled property technology firm, announced it raised $25M in funding.

Agentic AI startup CrewAI closes $18M funding round

With $13M in seed funding, DataCrunch wants to be Europe’s first AI cloud hyperscaler — powered by renewable energy (per TechCrunch)

Polish alternative to Snowflake and Databricks snaps up $11M for high-volume data analytics

Dunia raises $11.5M to accelerate electrochemical materials discovery

Hue raises $4.5M to expand AI-powered video platform for e-commerce

There’s more AI and big data news on SiliconANGLE

Around the enterprise: App store armageddon?

Money matters

Earnings:

IBM revenue misses, but execs say AI will drive future growth

ServiceNow delivers another earnings beat, but stock dips as guidance disappoints

SAP’s stock rises on fast-growing cloud revenue

Seagate stock falls despite beating earnings forecasts

Texas Instruments forecasts earnings below estimates on weak chip demand in industrial markets

Nvidia supplier SK Hynix posts record Q3 profit, beats forecasts

Western Digital stock jumps on better-than-expected earnings

Intel CEO Paul Otellini proposed buying Nvidia way back in 2005, but the board said no. Seems like Intel’s board has long been the problem, though in 2005, Nvidia wasn’t NVIDIA and $20 billion was a lot of money at the time.

A glimmer of hope for tech IPOs? Ingram Micro returns to public markets in $409M IPO

AI procurement platform provider Zip raises $190M, boosting valuation to $2.2B

Payment processing startup Finix closes $75M investment

Bluesky raises $15M to expand decentralized social network and developer ecosystem

In other enterprise news

This may turn out to be a seismic event or just a negotiating tactic: Arm moves to cancel Qualcomm’s instruction set architecture license

TSMC notifies US of potential attempt to make Huawei chip using its fabs

Qualcomm’s Snapdragon 8 Elite chipset features world’s fastest mobile CPU: This could be a big deal. As Qualcomm CEO Cristiano Amon told Axios: “You may pick the agents of your choice, and once the agents understand what you want, the agent is free and not bounded by the OS, an app store or anything like that.”

Former Intel directors call on the chipmaker to split its fab and design units But The Register’s Tobias Mann begs to disagree with a reasonable argument.

We have more news on cloud, infrastructure and apps

Cyber beat: Platform wars

Attack & response

I’d call this a security issue: Elon Musk has reportedly been in regular contact with Vladimir Putin

Breaking Analysis: How IBM and Palo Alto Networks team up to combat cyberthreats

Microsoft report: As US election nears, Russia, Iran and China step up influence efforts

IBM launches Guardium Data Security Center to address AI, quantum and hybrid cloud risks

SailPoint expands identity security platform with new cloud features, AI and automation

Google announces new AI-driven enhancements and security features for Chrome Enterprise and ChromeOS

Critical vulnerability in Fortinet’s FortiManager exploited in the wild

Money matters

Sophos to buy rival cybersecurity provider Secureworks for $859M

Socure to acquire Effectiv for $136M

Socket secures $40M to strengthen open-source software security

Cyber Guru raises $25M to expand AI-driven cybersecurity training platform

More cybersecurity news here

Elsewhere in tech: Waymo gets a huge new funding

Waymo raises $5.6B to grow its autonomous taxi service

Stripe acquires stablecoin infrastructure startup Bridge

Qualcomm and Google to collaborate on more intelligent cars with integrated generative AI

Apple and Goldman Sachs ordered to pay $89 million after Apple Card failures

Warehouse robot maker Nimble raises $106M Series C funding round at $1B valuation.

Tesla shares surge 11% on strong earnings and promise of more affordable models in 2025

Freeform raises $14M to build an AI-powered 3D printing factory

Meta will employ facial recognition technology to fight celebrity scam ads

And check out more news on emerging tech, blockchain and crypto and policy

Comings and goings: a high-profile Google Cloud departure

ServiceNow named Amit Zavery president, chief product officer and chief operating officer, leading product and engineering. He hails most recently from Google Cloud, and Oracle before that.

Dr. Ronnie Chatterji, a Duke University professor of business and public policy and former White House CHIPS coordinator, named OpenAI’s first chief economist. Meantime, longtime OpenAI policy researcher Miles Brundage, a senior adviser to its AGI readiness team, has left to focus on research and aid the nonprofit sector.

What’s next

Events

Oct. 28-30: TechCrunch Disrupt, San Francisco

Oct. 28-31: Constellation Research’s Connected Enterprise, Half Moon Bay, California

Earnings: a huge week coming for enterprise tech:

Monday, Oct. 28: F5 Networks

Tuesday, Oct. 29: Alphabet/Google, AMD, Check Point Software, Commvault and Snap

Wednesday, Oct. 30: Extreme Networks, Microsoft, Samsung, Twilio, Meta, Confluent, Informatica, Tenable, Coinbase and Robinhood

Thursday, Oct. 31: Uber, Amazon, Apple, Intel, Mobileye, Atlassian and OpenText

Image: SiliconANGLE/Google Gemini