Google Has a New AI-Weather Model for Cyclones. Should Experts Trust It?
On Thursday, Google announced an advancement powered by artificial intelligence that could change the way we predict hurricanes. Weather Lab is an interactive website that shows live and historic AI weather models, including its latest tropical cyclone model, which includes hurricanes. It was developed by Google DeepMind, the company's London-based AI research lab.
The cyclone model can predict the formation, track, intensity, size and shape of the storm. And it can create 50 possible scenarios up to 15 days ahead.
A representative for Google did not immediately respond to a request for comment.
How to use Weather Lab
Weather Lab's website lets experts compare AI weather models to physics-based models from the European Centre for Medium-Range Weather Forecasts to get more cyclone information sooner.
If experts are able to predict the storm earlier than a physics model, the extra time could help you and experts prepare for the impact of storms, especially those that could be life-threatening.
The problem with physics-based models is that they don't track a cyclone's intensity or track as accurately as Google's AI-powered model. When experts look at both types of models, "they can better anticipate a cyclone's path and intensity," according to Google.
The lab is running a few AI weather models in real time, but has included two years of previous predictions on the website so other researchers and experts can evaluate the models Google Deep Mind is creating. The two years researched used are not specified.
Don't ditch other weather services
It's worth noting that the Weather Lab can be helpful for future hurricane seasons and maybe even this one.
Weather Lab accurately predicted the paths of two 2025 cyclones, Honde and Garance. Other storm paths were accurately predicted almost seven days in advance. Google DeepMind partnered with the US National Hurricane Center to confirm that their approach and outputs were correct to make predictions.
However, Weather Lab is a research tool and even the live predictions are not official warnings. The lab still recommends relying on your local or national weather service.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Medscape
25 minutes ago
- Medscape
Immune Modulation in MDS: Our Core Strategies Need to Change
MILAN — Immune-based treatments continue to fall short in myelodysplastic syndromes (MDS). Now, experts from the European Hematology Association's (EHA) Specialized Working Group on MDS here at the EHA 2025 Congress are reevaluating their core strategies, from clinical trial endpoints to patient selection and timing. The researchers are suggesting the need for a shift in how these therapies are developed and deployed, as new data suggests that current benchmarks might be missing meaningful responses and that immune dysfunction in MDS can evolve over time. MDS are a heterogeneous group of clonal myeloid stem and progenitor cell disorders marked by ineffective hematopoiesis, cytopenias, and a risk of progression to acute myeloid leukemia. Immune dysregulation is thought to be a key driver in MDS pathogenesis, raising hopes that modulating immune responses within the bone marrow niche could alter disease trajectory. Yet, despite numerous attempts, immune-based interventions have largely failed to yield meaningful clinical results. Choosing the Right Endpoints Currently, few therapies targeting inflammation and immune activation are approved for MDS — namely antithymocyte globulin, with or without cyclosporine A, lenalidomide, and luspatercept. However, a growing list of investigational agents is under study, including cytokine inhibitors, signaling pathway inhibitors, immune checkpoint inhibitors, bispecific and trispecific antibodies, and chimeric antigen receptor cell-based therapies. Anne Sophie Kubasch, MD Anne Sophie Kubasch, MD, a hematologist at the University Medical Hospital Leipzig, Germany, presented preliminary findings from two trials by the German MDS Study Group: the phase 2 CANFIRE trial of the IL-1Beta inhibitor canakinumab and the LUCAS trial of the IRAK4 inhibitor emavusertib. The CANFIRE trial enrolled 11 patients treated for 6 months, while LUCAS included 36 patients treated for 4 months. Both studies failed to meet their primary endpoint, hematologic improvement in erythroid cells per International Working Group (IWG) 2018 criteria, by the end of the treatment period. Despite this, Kubasch noted that several patients opted to continue treatment due to perceived clinical benefit, such as reduced fatigue or general improvement. Some also showed biomarker changes or delayed hematologic responses. These findings led her to question whether hematologic improvement in erythroid cells is the most appropriate efficacy endpoint for immunomodulatory agents. 'We can argue that current IWG endpoint criteria are too rigid for immune modulation trials,' she said. 'Late or gradual responses may go undetected.' Kubasch suggested that future trials should incorporate patient-reported outcomes and quality-of-life metrics as co-primary endpoints. 'We may focus on composite endpoints combining various parameters, such as hematologic trends, biomarker shifts, molecular remission or stability, and, of course, patient-reported benefits.' Selecting the Right Patients A second key question, Kubasch said, is whether the right patients are being selected for immunotherapy. She cited a recent transcriptomic analysis of bone marrow CD34+ cells from 183 MDS patients that identified two immune subtypes based on the expression of six genes — including IRAK4 , the target of emavusertib. These subtypes were called hyperactive (HIC) and moderate (MIC) immune clusters. The HIC group had greater activation of immune-related pathways and higher infiltration of natural killer cells and M1 macrophages. 'We might be treating the wrong immune subgroup with immunomodulators,' she said, hypothesizing that it's possible that only MIC patients were being enrolled in their trials, contributing to negative results. She proposed a stepwise stratification approach: Baseline assessment (low-risk MDS diagnosis, anemia, transfusion burden) Genetic stratification (cytogenetics, mutations, resistance markers) Immune transcriptomic profiling (HIC/MIC clustering) 'Selective inclusion of patients with immune activation, such as those with an HIC profile, is essential,' she concluded. Targeting the Right Immune Alterations Valeria Santini, MD, associate professor of hematology at the University of Florence Medical School, Italy, focused her remarks on immunosuppressive therapy, a long-used but inconsistently effective strategy in MDS. To re-evaluate its role, Santini reviewed older literature — what she called 'archaeological digging.' She showed that complete responses and survival benefits are achievable in some patients, albeit rarely. Importantly, she emphasized that immune dysfunction in MDS is dynamic. 'It changes during the course of the disease,' she said. Therefore, precise immune profiling at both the subtype and patient level is needed. Optimization of immunosuppressive therapy is hampered by the lack of prospective randomized trials, heterogeneity in study populations, poor characterization of immune abnormalities, and the need for integration with novel agents. 'It is not one cure for all,' she said. 'By selecting patients, we might be more successful than in the past.' Finding the Right Drugs and the Right Timing Lionel Adès, MD, PhD, professor of hematology at Saint-Louis Hospital, Paris, France, commented to Medscape Medical News that, while the biological rationale for immune modulation is strong, clinical evidence remains weak. Lionel Adès, MD, PhD 'Maybe we don't have the right drugs or combinations. Maybe we're using them too late in the disease course,' he said. Immune dysregulation might play a role early in MDS, but by the time patients are treated, it could be too late to alter the disease with immune-targeted agents. Moreover, Adès noted that immune therapies often require extended treatment periods before activity can be detected, while traditional MDS trial endpoints focus on earlier timepoints. 'If a drug appears completely ineffective, it's not ethical to keep a patient on trial for a year just to see if something happens,' he said. Still, he remains cautiously optimistic. He pointed to the EHA-sponsored i4MDS Consortium, which aims to deepen understanding of MDS immunobiology, and to data from the STIMULUS MDS 2 study presented during the session by Amer Zeidan, MBBS, MHS, of Yale University, New Haven, Connecticut. That study evaluated the TIM-3 immune checkpoint inhibitor sabatolimab plus azacitidine as a first-line therapy in MDS. Although the trial failed to show a survival advantage overall, subgroup analysis revealed a benefit among female patients. 'This is something very interesting,' Adès said. 'It's something we didn't look at before, because we never thought survival could differ between male and female MDS patients.' While sabatolimab development has since been discontinued, other agents targeting the same pathway remain in the pipeline. 'This finding could be valuable for stratifying future trials by sex,' he said. Kubasch reported relationships with BMS, Curis, Jazz, Johnson & Johnson, and Novartis. Santini reported serving on advisory boards for AbbVie, Ascentage, BMS, Geron, GSK, Keros, Novartis, Servier, and Takeda. Adès disclosed research support from BMS/Celgene and AbbVie, and advisory roles with AbbVie, Amgen, BMS, Jazz, Novartis, Roche, and Takeda.
Wall Street Journal
32 minutes ago
- Wall Street Journal
Advanced Micro Devices Sets Sights on Nvidia in AI Race
Advanced Micro Devices wants to go head-to-head with Nvidia. Wall Street is split on whether AMD will succeed. The chip maker at its Advancing AI event on Thursday said it can meet surging demand for artificial-intelligence hardware, positioning itself as a realistic alternative to Nvidia, which has emerged as one of the world's most valuable companies thanks to its dominance in AI chips.
Yahoo
38 minutes ago
- Yahoo
Nvidia, Amazon, Google Lead Tech Selloff After Israel Strikes Iran
June 13 - Tech stocks tracked broader markets lower as Middle East tensions rose. The iShares Expanded Tech-Software Sector ETF slipped about 1%, while the Philadelphia Semiconductor Index fell about 2%. Among large-cap names, Google (NASDAQ:GOOGL) and Amazon (NASDAQ:AMZN) each dropped about 1%, and Nvidia (NASDAQ:NVDA) slid about 1%. Meta Platforms (NASDAQ:META) was little changed, and Microsoft (NASDAQ:MSFT) also saw minimal movement. Apple (NASDAQ:AAPL) showed modest gains but remained within a narrow range. Palantir (NASDAQ:PLTR) bucked the trend, climbing about 1%, possibly on its ties with defense contracts. (NASDAQ:MNDY) declined about 3%, reflecting sensitivity to geopolitical uncertainty. Semiconductor stocks broadly ceded ground: Taiwan Semiconductor Manufacturing (TSM) fell about 2%, Micron Technology (NASDAQ:MU) slipped about 1%, Qualcomm (NASDAQ:QCOM) lost about 1%, and Marvell Technology (MRVL) slid about 2%. IBM (NYSE:IBM) and Kyndryl Holdings (NYSE:KD) each edged down about 1%. The S&P 500, Nasdaq Composite, and Dow all retreated roughly 1% as investors weighed the impact of Israel's strikes on Iran. The effects of the attack have cascaded across global markets, with a strong risk-off move for several asset classes, said Deutsche Bank's Henry Allen. Israeli Prime Minister Benjamin Netanyahu said, This operation will continue for as many days as it takes to remove this threat, underscoring uncertainty. Market participants remain alert to how the situation might affect tech demand and broader sentiment. This article first appeared on GuruFocus. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
