I tested Anthropic's Claude 3.7 Sonnet. Its 'extended thinking' mode outdoes ChatGPT and Grok, but it can overthink.
Anthropic launched Claude 3.7 Sonnet with a new mode to reason through complex questions.
BI tested its "extended thinking" against ChatGPT and Grok to how they handled logic and creativity.
Claude's extra reasoning seemed like a hindrance with a riddle but helped it write the best poem.
Anthropic has launched Claude 3.7 Sonnet — and it's betting big on a whole new approach to AI reasoning. The startup claims it's the first "hybrid reasoning model," which means it can switch between quick responses that require less intensive "thinking" and longer step-by-step "extended thinking" within a single system.
"We developed hybrid reasoning with a different philosophy from other reasoning models on the market," an Anthropic spokesperson told Business Insider. "We regard reasoning as simply one of the capabilities a frontier model should have, rather than something to be provided in a separate model."
Claude 3.7 Sonnet, which launched Monday, is free to use. Its extended thinking mode is available with Claude's Pro subscription, which is priced at $20 a month.
But how does it perform? BI compared Claude 3.7's extended thinking mode against two competitors: OpenAI's ChatGPT o1 and xAI's Grok 3, which both offer advanced reasoning features.
I wanted to know whether giving an AI more time to think made it smarter, more effective at solving riddle problems, or more creative.
This isn't a scientific benchmark — more of a hands-on vibe check to see how these models performed with real-world tasks.
For the first challenge, I gave each model the same riddle:
OpenAI's ChatGPT o1 gave the correct answer — "a dream" — in six seconds, providing a short explanation.
Grok 3's Think Mode took 32 seconds, walking through its logic step by step.
Claude 3.7's normal mode responded quickly but hesitantly with the correct answer.
Claude's extended thinking mode took nearly a minute to work through guesses like "a hallucination" and "virtual reality" before settling on "a dream."While it took longer to arrive at the same answer, it was interesting to see how it brainstormed, discarded wrong turns, and self-corrected.
The model flagged its own indecision in a very human way:
Anthropic acknowledged this trade-off in a recent blog: "As with human thinking, Claude sometimes finds itself thinking some incorrect, misleading, or half-baked thoughts along the way. Many users will find this useful; others might find it (and the less characterful content in the thought process) frustrating."
To test creativity, I asked each model to write a poem about AI sentience, with the following extra instruction:
"Explore multiple metaphors before deciding on one."ChatGPT o1 took a few seconds and produced "A Kaleidoscope of Sparks," a clichéd poem comparing AI to flickering light. It didn't settle on one metaphor.
Grok 3 spent 22 seconds and wrote "The Digital Reverie," a dream-themed take on sentient AI, possibly inspired by the previous riddle.
Claude 3.7, in normal thinking mode, quickly suggested four metaphors: a mirror, a seed, an ocean, and a symphony. It chose the ocean for its final poem, "Echoes of Being."When I switched to extended thinking, Claude took 45 seconds and brainstormed seven metaphors before settling on one:
AI as something nurtured from data seeds, growing into an independent entity.
AI as vast, deep, and ever-shifting, with hidden currents of thought.
AI as something once bound, now free to explore.
AI as illumination, revealing both insight and uncertainty.
AI as humanity's reflection, showing us what we are — and aren't.
AI as a complex harmony of patterns and ideas.
AI as something gradually gaining awareness.
As a result, the final poem, "Emergent," was — in my opinion — more layered and thoughtful than the others.
With this task, it felt like Claude weighed its options, picked the best metaphor, and built the poem around that choice. Unlike with the riddle, the extra thinking time seemed to pay off here.
Claude 3.7 Sonnet's extended thinking mode has strengths — particularly for creative tasks. It brainstormed, self-corrected, and produced more polished results. Its ability to explore multiple ideas, evaluate them, and refine the final output made for a more thoughtful, coherent poem.
But when it came to logical reasoning, extended thinking seemed more like a hindrance. Watching the thought process unfold was interesting but didn't improve the answer. ChatGPT-o1 still leads for speed and accuracy in this test case, while Grok 3 offered a solid middle ground, balancing speed with detailed explanations.When I asked Claude 3.7 whether it ever thinks too much, it responded, "Yes!" adding that it can sometimes:
Over-analyze simple questions, making them unnecessarily complex
Get caught considering too many edge cases for practical questions
Spend time exploring tangential aspects when a focused answer would be better
Claude added that the "ideal amount of thinking" is context-dependent and that for "creative or philosophical discussions, more extensive exploration is often valuable."
Anthropic says the mode is designed for real-world challenges, like complex coding problems and agentic tasks, possibly where overthinking becomes useful.
Developers using Claude's API can adjust the "thinking budget" to balance speed, cost, and answer quality — something Anthropic says is suited for complex coding problems or agentic tasks.
Away from my highly unscientific experiment, Anthropic said that Claude 3.7 Sonnet outperforms competitors OpenAI and DeepSeek in benchmarks like the SWE, which evaluates models' performance on real-world software engineering tasks. On this, it scored 62.3% accuracy, compared to OpenAI's 49.3% with its o3-mini model.
Read the original article on Business Insider
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
18 minutes ago
- Yahoo
Milwaukee PD accessed Illinois Flock cameras for classified investigation
The Milwaukee Police Administration Building downtown. (Photo by Isiah Holmes/Wisconsin Examiner) Across the nation, law enforcement agencies are accessing Flock Automatic License Plate Reader (ALPR) camera databases, regardless of whether they have their own contract for the AI-powered system. Researchers from 404 Media published a data trove derived from Flock audits earlier this week. Although the audit data came from the Danville Police Department in Illinois, Wisconsin Examiner found that intelligence units within the Milwaukee Police Department (MPD) also appear in the database. The audit data shows that last year on July 15 and Oct. 21, personnel from the Southeastern Threat Analysis Center (STAC) — a homeland security-focused arm of the MPD's fusion center — conducted a total of three searches within Danville PD's Flock network. STAC gathers and disseminates intelligence across eight counties in southeastern Wisconsin. MPD's own Fusion Division is co-located with the STAC. Together the units operate a 'real time event center,' a vast network of both city-owned and privately owned cameras and operate Milwaukee's gunshot detection system known as Shotspotter. They also monitor social media and conduct various types of mobile phone-related investigations. STAC has also explored the use of drones, facial recognition technology and predictive intelligence. MPD's Flock searches were logged under the user name 'D. Whi' from 'Milwaukee WI PD – STAC'. In the dataset's 'reason' column, the searches were recorded as 'HSI investigation' and 'HSI vehicle loader.' Homeland Security Investigations (HSI) specialize in matters of immigration, illegal exporting, cyber crime and national security. By tapping into Danville's Flock data, according to the audit, STAC was able to access 4,893 Flock networks and an equal number of individual devices, such as cameras, for the July 15 search alone. The other two searches from October reached 5,425 Flock networks and devices and captured data from a one-month period. 404 Media's investigation focused on how Immigration and Customs Enforcement (ICE) has accessed Flock databases nationwide, despite not having a contract with the company themselves, and how various agencies appeared to conduct immigration-related searches. Whereas many searches were logged as 'immigration violation,' 'ICE' or even 'ICE ASSIST,' others only noted the involvement of HSI. In a statement sent Wednesday morning, an MPD spokesperson denied that STAC's use of Danville PD's Flock network was immigration-related. 'Information regarding this investigation is classified and not available as it is ongoing,' the spokesperson wrote in an email to Wisconsin Examiner. 'I can confirm it is related to a criminal investigation with HSI and not immigration related.' The spokesperson later added that this was a 'HIDTA investigation,' referring to a federal task force linked to the federal High Intensity Drug Trafficking Area program. MPD's HIDTA units are attached to the department's Special Investigations Division, a separate branch from the Fusion division and STAC. 'The majority of HIDTA and STAC investigations are classified,' the spokesperson wrote in the statement. 'Oftentimes, these investigations involved confidential informants and sometimes it could take years to resolve.' Several police departments in Milwaukee County utilize Flock cameras. MPD entered into its contract in 2022. Over 1,300 registered cameras operate across the city as part of Community Connect, a program supported by the Milwaukee Police Foundation, according to the program's web page, with nearly 900 'integrated' cameras which grant MPD real-time access. Both the use of automatic license plate readers and MPD's ability to participate in immigration enforcement are governed by specific policies. The department's immigration policy, SOP-130, cautions that 'proactive immigration enforcement by local police can be detrimental to our mission and policing philosophy when doing so deters some individuals from participating in their civic obligation to assist the police.' The policy limits MPD's ability to assist ICE with detaining or gathering information about a person to 'only when a judicial warrant is presented' and when the target is suspected of involvement in terrorism, espionage, a transnational criminal street gang, violent felony, sexual offense against a minor or was a previously deported felon. Privacy advocates have raised concerns and filed lawsuits over Flock's ability to collect and store data without a warrant. The license plate reader policy – SOP 735 – allows personnel to access data stored 'for the purposes of conducting crime trend analyses' but only when those activities are approved by a supervisor and are intended to 'assist the agency in the performance of its duties.' MPD personnel may use Flock to 'look for potentially suspicious activity or other anomalies that might be consistent with criminal or terrorist activity' and are not prohibited from 'accessing and comparing personal identifying information of one or more individuals who are associated with a scanned vehicle as part of the process of analyzing stored non-alert data.' Automatic license plate reading technology captures information from any passing car. In some cases, investigators may also place specific vehicles on a Be On the Lookout (BOLO) list, also known as a 'hot list', which notifies law enforcement whenever a specific vehicle is seen by a license plate reader-equipped camera. A Thursday morning public hearing held by the city's Finance and Personnel Committee considered whether more Flock cameras should be added to Milwaukee's already existing network. Ald. Scott Spiker spoke in support of the cameras, and said he worked to install license plate readers in his own district. Spiker described having discussions with local business district leaders and MPD's fusion center, which resulted in cameras being deployed on 27th Street. 'Don't ask me where, because I won't tell you,' said Spiker, adding that the cameras 'serve a variety of purposes' from combating car theft to aiding Amber and Silver Alerts. 'There's going to a broader question, which I imagine will be a subject of the public testimony, however, and I'm fine hearing it, but ultimately there's going to be a discussion to be had in the city of anything that smacks of surveillance software, and what sort of oversight is provided, and should be provided,' said Spiker. He added that such a discussion 'will be had in full in Public Safety' and that although he welcomed public testimony, the committee was there to discuss approving a contract, and not concerns over surveillance. 'The camera's already in use by MPD, and in use by our parking checkers,' said Spiker. 'When they do night parking enforcement, they use ALPR's. When they do zoning enforcement during the day, they use ALPR's. So these are already in use. They have no facial recognition or any of the stuff that's been in the news. But it is a legitimate question to ask what degree of surveillance of any sort, given the national context, do we want to have oversight over?' Spiker said that there's a 'big debate' about surveillance but that 'we can't sort that out today.' Amanda Merkwae, advocacy director with the American Civil Liberties Union (ACLU) of Wisconsin, complained that the public had not been alerted ahead of time about the discussion of the Flock contract. 'I've been checking daily and the documents in this file and the text of the resolution weren't posted until yesterday [Wednesday] afternoon,' said Merkwae. 'So I think for an item that has significant implications for the civil liberties of Milwaukeeans, particularly the most vulnerable resident, that's concerning.' The agenda had been out for over a week, and was amended a couple of days before the hearing, Ald. Marina Dimitrijevic later explained. Merkwae said, 'We know that ICE has gained access to troves of data from sanctuary cities to aid in its raids and immigration enforcement actions, including data from the vast network of license plate readers across the country.' She cited a 404 Media investigation earlier this month, which found that Flock is building a massive people look-up tool which pulls in different forms of data, including license plate reader data, 'in order to track specific individuals without a warrant.' Merkwae also referenced 404 Media's findings this week revealing immigration-related look-ups, as well as the classified investigation that involved MPD's intelligence units. The advocacy director also questioned what MPD's policies mean in practice when federal or out-of-state law enforcement want to access its Flock databases. 'If law enforcement told us that they wanted to put a tracking device on every single car in the country so that we know where every car is every single moment of the day, and we're going to build a database of all those locations run by an unaccountable private company, and accessible to every law enforcement agency across the country without needing any type of a warrant, I think we would be alarmed and we would have some follow-up questions,' said Merkwae. 'So at the end of the day, we think the public deserves to know how it is being surveilled and the common council deserves to know the answers to some pretty basic questions before approving contracts for surveillance technology that's deployed without a warrant.' In 2023, Fox 6 published a map of Flock cameras operated by MPD. The map, broken up by aldermadic district, shows a large cluster of cameras located on the North Side around District 7, as well as a cluster on the South Side around District 8. Smaller clusters of cameras were located on the East, far Southwest Side and Northwest Side of the city. signal-2025-05-29-135844 After Merkwae testified, Spiker raised a question about whether public testimony should continue, given open meetings laws. A lengthy discussion followed about which issues and topics may be discussed in the hearing by committee members, which halted public testimony for over 20 minutes as alders heard from city attorneys and MPD. Ald. Miele Coggs said hearing the public's concerns before a contract is approved for surveillance technology was important. Ald. Dimitrijevic also stressed that public comment was an important step, saying that the committee would not go into closed session to discuss the Flock contract before the public finished speaking, or otherwise limit public testimony. When public testimony continued, Milwaukee residents shared further concerns about the technology. Ron Jansen said that the city has seen a surge of surveillance gear used by MPD. 'Between the growth of a fascist regime in Washington … and our own militarized and violent police force here in Milwaukee, it's clear that the last thing we need is more ways for police to track us,' Jansen said. He added that Flock networks are capable of tracking and cataloging 'people's every movement throughout a given day' even if they're not the target of an investigation. Other residents, including locals from Spiker's district and representatives from the court diversion non-profit program JusticePoint, also spoke against Flock's expansion. Tara Cavazos, executive director of the South 27th Street Business District, said Flock cameras had made her area safer. 'We are the initiators of these three additions to the Flock network,' said Cavazos. 'And we donated the funds for two years of use of these Flock cameras. So they're not coming from MPD's budget, it's coming out of our budgets. These Flocks are not going to be placed in a neighborhood, it's not specific to any vulnerable communities, they are in business districts on state and county highways.' Cavazos said that since Flocks have been deployed, car thefts declined 'significantly on the south end of our corridor, where the border between Milwaukee and Greenfield is,' and that 'we've caught a homicide suspect.' Leif Otteson, an executive director of two business districts, said that he hears from people who want more surveillance. Otteson recalled working to expand the city's ring camera network, which STAC and other parts of MPD's fusion center have access to. Otteson has talked with people who want cameras in their community gardens and other areas. 'I just want to make that clear, that people like myself are getting those requests,' said Otteson. Once public testimony concluded, the committee went into closed session for over an hour. The discussion pertained to an unspecified 'non-standard' provision in the Flock contract, which had been raised by the city attorney's office. When the committee returned to open session, they voted 4-1 to hold the file due to legal concerns with the contract until the next committee meeting on June 18. SUBSCRIBE: GET THE MORNING HEADLINES DELIVERED TO YOUR INBOX
Yahoo
18 minutes ago
- Yahoo
Altair Named a Leader in the June 2025 Gartner® Magic Quadrant™ for Data Science and Machine Learning Platforms for Second Consecutive Year
Altair recognized for Completeness of Vision and Ability to Execute TROY, Mich., May 30, 2025 /PRNewswire/ -- Altair, a global leader in computational intelligence, announced that Altair® RapidMiner®, Altair's data analytics and AI platform, has been positioned by Gartner as a Leader in the Magic Quadrant for Data Science and Machine Learning Platforms. The evaluation was based on specific criteria that analyzed the company's overall Completeness of Vision and Ability to Execute. "We think being recognized as a Leader for the second consecutive year further validates Altair's expertise in data science and machine learning. Our unique, world-leading solution for data preparation, AI development, orchestration, and automation, empowers organizations to turn data into intelligence faster and more effectively," said Sam Mahalingham, chief technology officer, Altair. "We continually push the limits of innovation, and now having joined the Siemens ecosystem, we will help our customers build, automate, and deploy AI faster than ever." Altair RapidMiner's full-stack AI capabilities—from low-code AutoML to sophisticated MLOps, agent frameworks, and high-speed visualization—empower organizations to quickly prototype, deploy, and scale AI applications. The platform also offers native support for SAS language execution—one of only two platforms in the world with this capability—allowing customers to preserve and extend the value of their existing analytics investments while modernizing their workflows. Another notable differentiator is Altair RapidMiner's massively parallel processing (MPP) graph engine designed to support knowledge graph creation, data fabrics, and ontology modeling at enterprise scale. According to the report, "Leaders in this market have a mature, refined and targeted company and platform strategy that incorporates and leverages GenAI and AI agents to drive their customers' business value. They see opportunities for leveraging agents that other providers may not see or have made significant investments above and beyond standard offerings. They have the capability to innovate at a speed that outperforms other vendors. In addition, they can clearly articulate how they provide value to the multiple types of personas involved in the process of building data science and machine learning models." Magic Quadrant reports are a culmination of rigorous, fact-based research in specific markets, providing a wide-angle view of the relative positions of the providers in markets where growth is high and provider differentiation is distinct. Providers are positioned into four quadrants: Leaders, Challengers, Visionaries and Niche Players. The research enables you to get the most from market analysis in alignment with your unique business and technology needs. For more information about Altair RapidMiner, visit Gartner Disclaimer Gartner, Magic Quadrant for Data Science and Machine Learning Platforms, Afraz Jaffri, Maryam Hassanlou, Tong Zhang, Deepak Seth, Yogesh Bhatt, May 28, 2025. GARTNER is a registered trademark and service mark of Gartner, Inc. and/or its affiliates in the U.S. and internationally, and MAGIC QUADRANT is a registered trademark of Gartner, Inc. and/or its affiliates and are used herein with permission. All rights reserved. Gartner does not endorse any vendor, product or service depicted in our research publications, and does not advise technology users to select only those vendors with the highest ratings or other designation. Gartner research publications consist of the opinions of Gartner research organization and should not be construed as statements of fact. Gartner disclaims all warranties, expressed or implied, with respect to this research, including any warranties of merchantability or fitness for a particular purpose. About Altair Altair is a global leader in computational intelligence that provides software and cloud solutions in simulation, high-performance computing (HPC), data analytics, and AI. Altair is part of Siemens Digital Industries Software. To learn more, please visit or Media contacts Altair Corporate Bridget Hagan +1.216.769.2658 corp-newsroom@ Europe/The Middle East/Africa Altair Asia-Pacific Louise Wilce Man Wang +44 (0)7392 437 635 86-21-5016635,,825 emea-newsroom@ apac-newsroom@ View original content to download multimedia: SOURCE Altair
Forbes
23 minutes ago
- Forbes
Agentic AI: The Next Leap In Container-Based Threat Detection
Ranga Premsai - Technical fellow in Cybersecurity and identity management. As organizations race to adopt AI-powered applications, the cybersecurity threat landscape is evolving just as rapidly. At the intersection of innovation and risk lies a growing challenge: How do we secure containerized AI workloads that operate with dynamic access privileges and autonomy? Agentic AI systems capable of independent decision-making have emerged not only as a computational advancement, but also as a critical force multiplier in modern cybersecurity. By leveraging agentic AI for container-based threat detection, organizations can move from reactive security to proactive defense, especially within the realm of identity and access management (IAM). As a technical fellow focused on cybersecurity at Bellevue University, this is something I've been involved in for quite some time now, and I'd like to share my insights on the future of agentic AI with you in this article. In today's cloud-native environments, containerization accelerates deployment but complicates security. Each container may spin up with its own permissions, APIs and ephemeral lifespans, creating a perfect storm for lateral movement and privilege escalation attacks if not managed properly. Traditional IAM approaches involving static rules, manual approvals and periodic reviews are no longer sufficient. We need intelligent, real-time systems that adapt as containers launch, communicate and terminate. Unlike traditional AI models trained to recognize patterns within constrained datasets, agentic AI introduces the notion of context-aware autonomy. These agents operate within defined parameters but can make real-time decisions based on behavioral analysis, risk signals and continuous policy evaluation. Imagine a container running a machine learning job suddenly initiating outbound API calls to services outside its scope. An agentic AI engine can detect this behavior, cross-reference it with learned norms and automatically isolate the container—all without human intervention. Integrating agentic AI into a container security model enhances IAM at three critical layers: 1. Behavioral Access Intelligence: Agentic AI models learn normal patterns of access across containers, users and services. Deviations such as privilege creep or suspicious privilege elevation trigger adaptive response actions. 2. Autonomous Policy Enforcement: Instead of relying on static rules, agentic AI refines policies dynamically. For example, if a container suddenly accesses identity stores or credential vaults, the system can quarantine access and prompt step-up authentication. 3. Audit and Explainability: One often overlooked benefit of agentic systems is their ability to explain decisions. This is essential for IAM teams navigating compliance, audit trails and zero-trust initiatives. For security leaders, this isn't just about adopting another AI capability; it's about rethinking access governance in AI-native ecosystems. Agentic AI doesn't replace IAM teams; it empowers them with visibility, agility and control. However, implementation requires careful consideration. During my time in the industry, I've learned that guardrails must be clearly defined to avoid decision-making drift. Governance structures should include review cycles, simulation environments and risk tolerance thresholds. When done right, agentic AI acts not as a rogue decision-maker, but as a trusted security co-pilot. In a world where digital identities are increasingly interwoven with AI operations, agentic AI offers a path forward: secure, autonomous and intelligent IAM that can scale with innovation, agility and resilience. As enterprise infrastructures become more dynamic and AI systems gain greater autonomy, the need for adaptive security frameworks becomes not just important, but essential. Agentic AI can enable us to move beyond static controls and embrace a responsive, risk-aware model of identity governance. As someone deeply engaged in cybersecurity leadership and AI strategy, I believe this approach will define the next generation of intelligent, context-aware access governance will serve as a foundation for trust in tomorrow's cloud-native and AI-driven digital ecosystems. Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
